Face recognition is a challenging task due to the complexity of pose variations, occlusion and the variety of face expressions performed by distinct subjects. Thus, many features have been proposed, however each feature has its own drawbacks. Therefore, in this paper, we propose a robust model called Krawtchouk moments convolutional neural networks (KMCNN) for face recognition. Our model is divided into two main steps. Firstly, we use 2D discrete orthogonal Krawtchouk moments to represent features. Then, we fed it into convolutional neural networks (CNN) for classification. The main goal of the proposed approach is to improve the classification accuracy of noisy grayscale face images. In fact, Krawtchouk moments are less sensitive to noisy effects. Moreover, they can extract pertinent features from an image using only low orders. To investigate the robustness of the proposed approach, two types of noise (salt and pepper and speckle) are added to three datasets (YaleB extended, our database of faces (ORL), and a subset of labeled faces in the wild (LFW)). Experimental results show that KMCNN is flexible and performs significantly better than using just CNN or when we combine it with other discrete moments such as Tchebichef, Hahn, Racah moments in most densities of noises.
A face recognition system using convolutional feature extraction with linear...IJECEIAES
Face recognition is one of the important biometric authentication research areas for security purposes in many fields such as pattern recognition and image processing. However, the human face recognitions have the major problem in machine learning and deep learning techniques, since input images vary with poses of people, different lighting conditions, various expressions, ages as well as illumination conditions and it makes the face recognition process poor in accuracy. In the present research, the resolution of the image patches is reduced by the max pooling layer in convolutional neural network (CNN) and also used to make the model robust than other traditional feature extraction technique called local multiple pattern (LMP). The extracted features are fed into the linear collaborative discriminant regression classification (LCDRC) for final face recognition. Due to optimization using CNN in LCDRC, the distance ratio between the classes has maximized and the distance of the features inside the class reduces. The results stated that the CNN-LCDRC achieved 93.10% and 87.60% of mean recognition accuracy, where traditional LCDRC achieved 83.35% and 77.70% of mean recognition accuracy on ORL and YALE databases respectively for the training number 8 (i.e. 80% of training and 20% of testing data).
Gender classification using custom convolutional neural networks architecture IJECEIAES
Gender classification demonstrates high accuracy in many previous works. However, it does not generalize very well in unconstrained settings and environments. Furthermore, many proposed convolutional neural network (CNN) based solutions vary significantly in their characteristics and architectures, which calls for optimal CNN architecture for this specific task. In this work, a hand-crafted, custom CNN architecture is proposed to distinguish between male and female facial images. This custom CNN requires smaller input image resolutions and significantly fewer trainable parameters than some popular state-of-the-arts such as GoogleNet and AlexNet. It also employs batch normalization layers which results in better computation efficiency. Based on experiments using publicly available datasets such as LFW, CelebA and IMDB-WIKI datasets, the proposed custom CNN delivered the fastest inference time in all tests, where it needs only 0.92ms to classify 1200 images on GPU, 1.79ms on CPU, and 2.51ms on VPU. The custom CNN also delivers performance on-par with state-ofthe-arts and even surpassed these methods in CelebA gender classification where it delivered the best result at 96% accuracy. Moreover, in a more challenging cross-dataset inference, custom CNN trained using CelebA dataset gives the best gender classification accuracy for tests on IMDB and WIKI datasets at 97% and 96% accuracy respectively.
A hybrid approach for face recognition using a convolutional neural network c...IAESIJAI
Facial recognition technology has been used in many fields such as security,
biometric identification, robotics, video surveillance, health, and commerce
due to its ease of implementation and minimal data processing time.
However, this technology is influenced by the presence of variations such as
pose, lighting, or occlusion. In this paper, we propose a new approach to
improve the accuracy rate of face recognition in the presence of variation or
occlusion, by combining feature extraction with a histogram of oriented
gradient (HOG), scale invariant feature transform (SIFT), Gabor, and the
Canny contour detector techniques, as well as a convolutional neural
network (CNN) architecture, tested with several combinations of the
activation function used (Softmax and Segmoïd) and the optimization
algorithm used during training (adam, Adamax, RMSprop, and stochastic
gradient descent (SGD)). For this, a preprocessing was performed on two
databases of our database of faces (ORL) and Sheffield faces used, then we
perform a feature extraction operation with the mentioned techniques and
then pass them to our used CNN architecture. The results of our simulations
show a high performance of the SIFT+CNN combination, in the case of the
presence of variations with an accuracy rate up to 100%.
Facial recognition based on enhanced neural networkIAESIJAI
Accurate automatic face recognition (FR) has only become a practical goal of biometrics research in recent years. Detection and recognition are the primary steps for identifying faces in this research, and The Viola-Jones algorithm implements to discover faces in images. This paper presents a neural network solution called modify bidirectional associative memory (MBAM). The basic idea is to recognize the image of a human's face, extract the face image, enter it into the MBAM, and identify it. The output ID for the face image from the network should be similar to the ID for the image entered previously in the training phase. The tests have conducted using the suggested model using 100 images. Results show that FR accuracy is 100% for all images used, and the accuracy after adding noise is the proportions that differ between the images used according to the noise ratio. Recognition results for the mobile camera images were more satisfactory than those for the Face94 dataset.
TOP 5 Most View Article From Academia in 2019sipij
TOP 5 Most View Article From Academia in 2019
Signal & Image Processing : An International Journal (SIPIJ)
ISSN : 0976 - 710X (Online) ; 2229 - 3922 (print)
http://www.airccse.org/journal/sipij/index.html
Selective local binary pattern with convolutional neural network for facial ...IJECEIAES
Variation in images in terms of head pose and illumination is a challenge in facial expression recognition. This research presents a hybrid approach that combines the conventional and deep learning, to improve facial expression recognition performance and aims to solve the challenge. We propose a selective local binary pattern (SLBP) method to obtain a more stable image representation fed to the learning process in convolutional neural network (CNN). In the preprocessing stage, we use adaptive gamma transformation to reduce illumination variability. The proposed SLBP selects the discriminant features in facial images with head pose variation using the median-based standard deviation of local binary pattern images. We experimented on the Karolinska directed emotional faces (KDEF) dataset containing thousands of images with variations in head pose and illumination and Japanese female facial expression (JAFFE) dataset containing seven facial expressions of Japanese females’ frontal faces. The experiments show that the proposed method is superior compared to the other related approaches with an accuracy of 92.21% on KDEF dataset and 94.28% on JAFFE dataset.
A face recognition system using convolutional feature extraction with linear...IJECEIAES
Face recognition is one of the important biometric authentication research areas for security purposes in many fields such as pattern recognition and image processing. However, the human face recognitions have the major problem in machine learning and deep learning techniques, since input images vary with poses of people, different lighting conditions, various expressions, ages as well as illumination conditions and it makes the face recognition process poor in accuracy. In the present research, the resolution of the image patches is reduced by the max pooling layer in convolutional neural network (CNN) and also used to make the model robust than other traditional feature extraction technique called local multiple pattern (LMP). The extracted features are fed into the linear collaborative discriminant regression classification (LCDRC) for final face recognition. Due to optimization using CNN in LCDRC, the distance ratio between the classes has maximized and the distance of the features inside the class reduces. The results stated that the CNN-LCDRC achieved 93.10% and 87.60% of mean recognition accuracy, where traditional LCDRC achieved 83.35% and 77.70% of mean recognition accuracy on ORL and YALE databases respectively for the training number 8 (i.e. 80% of training and 20% of testing data).
Gender classification using custom convolutional neural networks architecture IJECEIAES
Gender classification demonstrates high accuracy in many previous works. However, it does not generalize very well in unconstrained settings and environments. Furthermore, many proposed convolutional neural network (CNN) based solutions vary significantly in their characteristics and architectures, which calls for optimal CNN architecture for this specific task. In this work, a hand-crafted, custom CNN architecture is proposed to distinguish between male and female facial images. This custom CNN requires smaller input image resolutions and significantly fewer trainable parameters than some popular state-of-the-arts such as GoogleNet and AlexNet. It also employs batch normalization layers which results in better computation efficiency. Based on experiments using publicly available datasets such as LFW, CelebA and IMDB-WIKI datasets, the proposed custom CNN delivered the fastest inference time in all tests, where it needs only 0.92ms to classify 1200 images on GPU, 1.79ms on CPU, and 2.51ms on VPU. The custom CNN also delivers performance on-par with state-ofthe-arts and even surpassed these methods in CelebA gender classification where it delivered the best result at 96% accuracy. Moreover, in a more challenging cross-dataset inference, custom CNN trained using CelebA dataset gives the best gender classification accuracy for tests on IMDB and WIKI datasets at 97% and 96% accuracy respectively.
A hybrid approach for face recognition using a convolutional neural network c...IAESIJAI
Facial recognition technology has been used in many fields such as security,
biometric identification, robotics, video surveillance, health, and commerce
due to its ease of implementation and minimal data processing time.
However, this technology is influenced by the presence of variations such as
pose, lighting, or occlusion. In this paper, we propose a new approach to
improve the accuracy rate of face recognition in the presence of variation or
occlusion, by combining feature extraction with a histogram of oriented
gradient (HOG), scale invariant feature transform (SIFT), Gabor, and the
Canny contour detector techniques, as well as a convolutional neural
network (CNN) architecture, tested with several combinations of the
activation function used (Softmax and Segmoïd) and the optimization
algorithm used during training (adam, Adamax, RMSprop, and stochastic
gradient descent (SGD)). For this, a preprocessing was performed on two
databases of our database of faces (ORL) and Sheffield faces used, then we
perform a feature extraction operation with the mentioned techniques and
then pass them to our used CNN architecture. The results of our simulations
show a high performance of the SIFT+CNN combination, in the case of the
presence of variations with an accuracy rate up to 100%.
Facial recognition based on enhanced neural networkIAESIJAI
Accurate automatic face recognition (FR) has only become a practical goal of biometrics research in recent years. Detection and recognition are the primary steps for identifying faces in this research, and The Viola-Jones algorithm implements to discover faces in images. This paper presents a neural network solution called modify bidirectional associative memory (MBAM). The basic idea is to recognize the image of a human's face, extract the face image, enter it into the MBAM, and identify it. The output ID for the face image from the network should be similar to the ID for the image entered previously in the training phase. The tests have conducted using the suggested model using 100 images. Results show that FR accuracy is 100% for all images used, and the accuracy after adding noise is the proportions that differ between the images used according to the noise ratio. Recognition results for the mobile camera images were more satisfactory than those for the Face94 dataset.
TOP 5 Most View Article From Academia in 2019sipij
TOP 5 Most View Article From Academia in 2019
Signal & Image Processing : An International Journal (SIPIJ)
ISSN : 0976 - 710X (Online) ; 2229 - 3922 (print)
http://www.airccse.org/journal/sipij/index.html
Selective local binary pattern with convolutional neural network for facial ...IJECEIAES
Variation in images in terms of head pose and illumination is a challenge in facial expression recognition. This research presents a hybrid approach that combines the conventional and deep learning, to improve facial expression recognition performance and aims to solve the challenge. We propose a selective local binary pattern (SLBP) method to obtain a more stable image representation fed to the learning process in convolutional neural network (CNN). In the preprocessing stage, we use adaptive gamma transformation to reduce illumination variability. The proposed SLBP selects the discriminant features in facial images with head pose variation using the median-based standard deviation of local binary pattern images. We experimented on the Karolinska directed emotional faces (KDEF) dataset containing thousands of images with variations in head pose and illumination and Japanese female facial expression (JAFFE) dataset containing seven facial expressions of Japanese females’ frontal faces. The experiments show that the proposed method is superior compared to the other related approaches with an accuracy of 92.21% on KDEF dataset and 94.28% on JAFFE dataset.
Deep Neural Networks (DNNs) have shown to outperformtraditionalmethodsinvariousvisualrecognitiontasks including Facial Expression Recognition (FER). In spite of efforts made to improve the accuracy of FER systems using DNN, existing methods still are not generalizable enough in practical applications. This paper proposes a 3D Convolutional Neural Network method for FER in videos. This new network architecture consists of 3D Inception-ResNet layers followed by an LSTM unit that together extracts the spatial relations within facial images as well as the temporal relations between different frames in the video. Facial landmark points are also used as inputs to our network which emphasize on the importance of facial components rather than the facial regions that may not contribute significantly to generating facial expressions. Our proposed methodisevaluatedusingfourpubliclyavailabledatabases in subject-independent and cross-database tasks and outperforms state-of-the-art methods.
Deep Neural Networks (DNNs) have shown to outperformtraditionalmethodsinvariousvisualrecognitiontasks including Facial Expression Recognition (FER). In spite of efforts made to improve the accuracy of FER systems using DNN, existing methods still are not generalizable enough in practical applications. This paper proposes a 3D Convolutional Neural Network method for FER in videos. This new network architecture consists of 3D Inception-ResNet layers followed by an LSTM unit that together extracts the spatial relations within facial images as well as the temporal relations between different frames in the video. Facial landmark points are also used as inputs to our network which emphasize on the importance of facial components rather than the facial regions that may not contribute significantly to generating facial expressions. Our proposed methodisevaluatedusingfourpubliclyavailabledatabases in subject-independent and cross-database tasks and outperforms state-of-the-art methods.
Social distance and face mask detector system exploiting transfer learningIJECEIAES
As time advances, the use of deep learning-based object detection algorithms has also evolved leading to developments of new human-computer interactions, facilitating an exploration of various domains. Considering the automated process of detection, systems suitable for detecting violations are developed. One such applications is the social distancing and face mask detectors to control air-borne diseases. The objective of this research is to deploy transfer learning on object detection models for spotting violations in face masks and physical distance rules in real-time. The common drawbacks of existing models are low accuracy and inability to detect in real-time. The MobileNetV2 object detection model and YOLOv3 model with Euclidean distance measure have been used for detection of face mask and physical distancing. A proactive transfer learning approach is used to perform the functionality of face mask classification on the patterns obtained from the social distance detector model. On implementing the application on various surveillance footage, it was observed that the system could classify masked and unmasked faces and if social distancing was maintained or not with accuracies 99% and 94% respectively. The models exhibited high accuracy on testing and the system can be infused with the existing internet protocol (IP) cameras or surveillance systems for real-time surveillance of face masks and physical distancing rules effectively.
Facial emotion recognition using deep learning detector and classifier IJECEIAES
Numerous research works have been put forward over the years to advance the field of facial expression recognition which until today, is still considered a challenging task. The selection of image color space and the use of facial alignment as preprocessing steps may collectively pose a significant impact on the accuracy and computational cost of facial emotion recognition, which is crucial to optimize the speed-accuracy trade-off. This paper proposed a deep learning-based facial emotion recognition pipeline that can be used to predict the emotion of detected face regions in video sequences. Five well-known state-of-the-art convolutional neural network architectures are used for training the emotion classifier to identify the network architecture which gives the best speed-accuracy trade-off. Two distinct facial emotion training datasets are prepared to investigate the effect of image color space and facial alignment on the performance of facial emotion recognition. Experimental results show that training a facial expression recognition model with grayscale-aligned facial images is preferable as it offers better recognition rates with lower detection latency. The lightweight MobileNet_v1 is identified as the best-performing model with WM=0.75 and RM=160 as its hyperparameters, achieving an overall accuracy of 86.42% on the testing video dataset.
Deep learning based masked face recognition in the era of the COVID-19 pandemicIJECEIAES
During the coronavirus disease 2019 (COVID-19) pandemic, monitoring for wearing masks obtains a crucial attention due to the effect of wearing masks to prevent the spread of coronavirus. This work introduces two deep learning models, the former based on pre-trained convolutional neural network (CNN) which called MobileNetv2, and the latter is a new CNN architecture. These two models have been used to detect masked face with three classes (correct, not correct, and no mask). The experiments conducted on benchmark dataset which is face mask detection dataset from Kaggle. Moreover, the comparison between two models is driven to evaluate the results of these two proposed models.
COMPRESSION BASED FACE RECOGNITION USING DWT AND SVMsipij
The biometric is used to identify a person effectively and employ in almost all applications of day to day
activities. In this paper, we propose compression based face recognition using Discrete Wavelet Transform
(DWT) and Support Vector Machine (SVM). The novel concept of converting many images of single person
into one image using averaging technique is introduced to reduce execution time and memory. The DWT is
applied on averaged face image to obtain approximation (LL) and detailed bands. The LL band coefficients
are given as input to SVM to obtain Support vectors (SV’s). The LL coefficients of DWT and SV’s are fused
based on arithmetic addition to extract final features. The Euclidean Distance (ED) is used to compare test
image features with database image features to compute performance parameters. It is observed that, the
proposed algorithm is better in terms of performance compared to existing algorithms.
Enhanced Face Detection Based on Haar-Like and MB-LBP FeaturesDr. Amarjeet Singh
The effective real-time face detection framework
proposed by Viola and Jones gained much popularity due its
computational efficiency and its simplicity. A notable
variant replaces the original Haar-like features with MBLBP (Multi-Block Local Binary Pattern) which are defined
by the local binary pattern operator, both detector types are
integrated into the OpenCV library. However, each
descriptor and its evaluation method has its own set of
strengths and setbacks. In this paper, an enhanced two-layer
face detector composed of both Haar-like and MB-LBP
features is presented. Haar-like features are employed as a
coarse filter but with a new evaluation involving dual
threshold. The already established MB-LBPs are arranged
as the fine filter of the detector. The Gentle AdaBoost
learning algorithm is deployed for the training of the
proposed detector to reach the classification and
performance potential. Experiments show that in the early
stages of classification, Haar features with dual threshold
are more discriminative than MB-LBP and original Haarlike features with respect to number of features required
and computation. Benchmarking the proposed detector
demonstrate overall 12% higher detection rate at 17% false
alarm over using MB-LBP features singly while performing
with ×3 speedup.
In this paper, an attempt has been made to extract texture
features from facial images using an improved method of
Illumination Invariant Feature Descriptor. The proposed local
ternary Pattern based feature extractor viz., Steady Illumination
Local Ternary Pattern (SIcLTP) has been used to extract texture
features from Indian face database. The similarity matching
between two extracted feature sets has been obtained using Zero
Mean Sum of Squared Differences (ZSSD). The RGB facial images
are first converted into the YIQ colour space to reduce the
redundancy of the RGB images. The result obtained has been
analysed using Receiver Operating Characteristic curve, and is
found to be promising. Finally the results are validated with
standard local binary pattern (LBP) extractor.
FACE EXPRESSION RECOGNITION USING CONVOLUTION NEURAL NETWORK (CNN) MODELS ijgca
This paper proposes the design of a Facial Expression Recognition (FER) system based on deep
convolutional neural network by using three model. In this work, a simple solution for facial expression
recognition that uses a combination of algorithms for face detection, feature extraction and classification
is discussed. The proposed method uses CNN models with SVM classifier and evaluates them, these models
are Alex-net model, VGG-16 model and Res-Net model. Experiments are carried out on the Extended
Cohn-Kanada (CK+) datasets to determine the recognition accuracy for the proposed FER system. In this
study the accuracy of AlexNet model compared with Vgg16 model and ResNet model. The result show that
AlexNet model achieved the best accuracy (88.2%) compared to other models.
Possibility fuzzy c means clustering for expression invariant face recognitionIJCI JOURNAL
Face being the most natural method of identification for humans is one of the most significant biometric
modalities and various methods to achieve efficient face recognition have been proposed. However the
changes in face owing to different expressions, pose, makeup, illumination, age bring about marked
variations in the facial image. These changes will inevitably occur and they can be controlled only till a
certain degree beyond which they are bound to happen and will affect the face thereby adversely impacting
the performance of any face recognition system. This paper proposes a strategy to improve the
classification methodology in face recognition by using Possibility Fuzzy C-Means Clustering (PFCM).
This clustering technique was used for face recognition due to its properties like outlier insensitivity which
make it a suitable candidate for use in designing such robust applications.PFCM is a hybridization of
Possibilistic C-Means (PCM) and Fuzzy C-Means (FCM) clustering algorithms. PFCM is a robust
clustering technique and is especially significant for its noise insensitivity. It has also resolved the
coincident clusters problem which is faced by other clustering techniques. Therefore the technique can also
be used to increase the overall robustness of a face recognition system and thereby increase its invariance
and make it a reliably usable biometric modality.
Content-based image retrieval based on corel dataset using deep learningIAESIJAI
A popular technique for retrieving images from huge and unlabeled image databases are content-based-image-retrieval (CBIR). However, the traditional information retrieval techniques do not satisfy users in terms of time consumption and accuracy. Additionally, the number of images accessible to users are growing due to web development and transmission networks. As the result, huge digital image creation occurs in many places. Therefore, quick access to these huge image databases and retrieving images like a query image from these huge image collections provides significant challenges and the need for an effective technique. Feature extraction and similarity measurement are important for the performance of a CBIR technique. This work proposes a simple but efficient deep-learning framework based on convolutional-neural networks (CNN) for the feature extraction phase in CBIR. The proposed CNN aims to reduce the semantic gap between low-level and high-level features. The similarity measurements are used to compute the distance between the query and database image features. When retrieving the first 10 pictures, an experiment on the Corel-1K dataset showed that the average precision was 0.88 with Euclidean distance, which was a big step up from the state-of-the-art approaches.
Facial image retrieval on semantic features using adaptive mean genetic algor...TELKOMNIKA JOURNAL
The emergence of larger databases has made image retrieval techniques an essential component and has led to the development of more efficient image retrieval systems. Retrieval can either be content or text-based. In this paper, the focus is on the content-based image retrieval from the FGNET database. Input query images are subjected to several processing techniques in the database before computing the squared Euclidean distance (SED) between them. The images with the shortest Euclidean distance are considered as a match and are retrieved. The processing techniques involve the application of the median modified Weiner filter (MMWF), extraction of the low-level features using histogram-oriented gradients (HOG), discrete wavelet transform (DWT), GIST, and Local tetra pattern (LTrP). Finally, the features are selected using Adaptive Mean Genetic Algorithm (AMGA). In this study, the average PSNR value obtained after applying the Wiener filter was 45.29. The performance of the AMGA was evaluated based on its precision, F-measure, and recall, and the obtained average values were respectively 0.75, 0.692, and 0.66. The performance matrix of the AMGA was compared to those of particle swarm optimization algorithm (PSO) and genetic algorithm (GA) and found to perform better; thus, proving its efficiency.
Depth-DensePose: an efficient densely connected deep learning model for came...IJECEIAES
Camera/image-based localization is important for many emerging applications such as augmented reality (AR), mixed reality, robotics, and self-driving. Camera localization is the problem of estimating both camera position and orientation with respect to an object. Use cases for camera localization depend on two key factors: accuracy and speed (latency). Therefore, this paper proposes Depth-DensePose, an efficient deep learning model for 6-degrees-of-freedom (6-DoF) camera-based localization. The Depth-DensePose utilizes the advantages of both DenseNets and adapted depthwise separable convolution (DS-Conv) to build a deeper and more efficient network. The proposed model consists of iterative depth-dense blocks. Each depth dense block contains two adapted DS-Conv with two kernel sizes 3 and 5, which are useful to retain both low-level as well as high-level features. We evaluate the proposed Depth-DensePose on the Cambridge Landmarks dataset, which shows that the Depth-DensePose outperforms the performance of related deep learning models for camera based localization. Furthermore, extensive experiments were conducted which proven the adapted DS-Conv is more efficient than the standard convolution. Especially, in terms of memory and processing time which is important to real-time and mobile applications.
The International Journal of Engineering and Science (IJES)theijes
The International Journal of Engineering & Science is aimed at providing a platform for researchers, engineers, scientists, or educators to publish their original research results, to exchange new ideas, to disseminate information in innovative designs, engineering experiences and technological skills. It is also the Journal's objective to promote engineering and technology education. All papers submitted to the Journal will be blind peer-reviewed. Only original articles will be published.
The face recognition technique gave in this study utilizes a reconfigurable organization of paramount threshold logic cells and can be utilized in the optional layer of a pixel exhibit. The technology used today for face recognition is neither either new nor particularly ancient. Face recognition is primarily employed for security reasons. Real-time applications have seen rapid growth in the demanding and fascinating field of face recognition. In recent years, face recognition has been the subject of intensive research. This report presents an up-to-date review of key human facial recognition studies. We begin by providing a general introduction of face recognition and its uses. The most recent facial recognition methods are then reviewed in the literature.
Bibliometric analysis highlighting the role of women in addressing climate ch...IJECEIAES
Fossil fuel consumption increased quickly, contributing to climate change
that is evident in unusual flooding and draughts, and global warming. Over
the past ten years, women's involvement in society has grown dramatically,
and they succeeded in playing a noticeable role in reducing climate change.
A bibliometric analysis of data from the last ten years has been carried out to
examine the role of women in addressing the climate change. The analysis's
findings discussed the relevant to the sustainable development goals (SDGs),
particularly SDG 7 and SDG 13. The results considered contributions made
by women in the various sectors while taking geographic dispersion into
account. The bibliometric analysis delves into topics including women's
leadership in environmental groups, their involvement in policymaking, their
contributions to sustainable development projects, and the influence of
gender diversity on attempts to mitigate climate change. This study's results
highlight how women have influenced policies and actions related to climate
change, point out areas of research deficiency and recommendations on how
to increase role of the women in addressing the climate change and
achieving sustainability. To achieve more successful results, this initiative
aims to highlight the significance of gender equality and encourage
inclusivity in climate change decision-making processes.
Voltage and frequency control of microgrid in presence of micro-turbine inter...IJECEIAES
The active and reactive load changes have a significant impact on voltage
and frequency. In this paper, in order to stabilize the microgrid (MG) against
load variations in islanding mode, the active and reactive power of all
distributed generators (DGs), including energy storage (battery), diesel
generator, and micro-turbine, are controlled. The micro-turbine generator is
connected to MG through a three-phase to three-phase matrix converter, and
the droop control method is applied for controlling the voltage and
frequency of MG. In addition, a method is introduced for voltage and
frequency control of micro-turbines in the transition state from gridconnected mode to islanding mode. A novel switching strategy of the matrix
converter is used for converting the high-frequency output voltage of the
micro-turbine to the grid-side frequency of the utility system. Moreover,
using the switching strategy, the low-order harmonics in the output current
and voltage are not produced, and consequently, the size of the output filter
would be reduced. In fact, the suggested control strategy is load-independent
and has no frequency conversion restrictions. The proposed approach for
voltage and frequency regulation demonstrates exceptional performance and
favorable response across various load alteration scenarios. The suggested
strategy is examined in several scenarios in the MG test systems, and the
simulation results are addressed.
More Related Content
Similar to Robust face recognition using convolutional neural networks combined with Krawtchouk moments
Deep Neural Networks (DNNs) have shown to outperformtraditionalmethodsinvariousvisualrecognitiontasks including Facial Expression Recognition (FER). In spite of efforts made to improve the accuracy of FER systems using DNN, existing methods still are not generalizable enough in practical applications. This paper proposes a 3D Convolutional Neural Network method for FER in videos. This new network architecture consists of 3D Inception-ResNet layers followed by an LSTM unit that together extracts the spatial relations within facial images as well as the temporal relations between different frames in the video. Facial landmark points are also used as inputs to our network which emphasize on the importance of facial components rather than the facial regions that may not contribute significantly to generating facial expressions. Our proposed methodisevaluatedusingfourpubliclyavailabledatabases in subject-independent and cross-database tasks and outperforms state-of-the-art methods.
Deep Neural Networks (DNNs) have shown to outperformtraditionalmethodsinvariousvisualrecognitiontasks including Facial Expression Recognition (FER). In spite of efforts made to improve the accuracy of FER systems using DNN, existing methods still are not generalizable enough in practical applications. This paper proposes a 3D Convolutional Neural Network method for FER in videos. This new network architecture consists of 3D Inception-ResNet layers followed by an LSTM unit that together extracts the spatial relations within facial images as well as the temporal relations between different frames in the video. Facial landmark points are also used as inputs to our network which emphasize on the importance of facial components rather than the facial regions that may not contribute significantly to generating facial expressions. Our proposed methodisevaluatedusingfourpubliclyavailabledatabases in subject-independent and cross-database tasks and outperforms state-of-the-art methods.
Social distance and face mask detector system exploiting transfer learningIJECEIAES
As time advances, the use of deep learning-based object detection algorithms has also evolved leading to developments of new human-computer interactions, facilitating an exploration of various domains. Considering the automated process of detection, systems suitable for detecting violations are developed. One such applications is the social distancing and face mask detectors to control air-borne diseases. The objective of this research is to deploy transfer learning on object detection models for spotting violations in face masks and physical distance rules in real-time. The common drawbacks of existing models are low accuracy and inability to detect in real-time. The MobileNetV2 object detection model and YOLOv3 model with Euclidean distance measure have been used for detection of face mask and physical distancing. A proactive transfer learning approach is used to perform the functionality of face mask classification on the patterns obtained from the social distance detector model. On implementing the application on various surveillance footage, it was observed that the system could classify masked and unmasked faces and if social distancing was maintained or not with accuracies 99% and 94% respectively. The models exhibited high accuracy on testing and the system can be infused with the existing internet protocol (IP) cameras or surveillance systems for real-time surveillance of face masks and physical distancing rules effectively.
Facial emotion recognition using deep learning detector and classifier IJECEIAES
Numerous research works have been put forward over the years to advance the field of facial expression recognition which until today, is still considered a challenging task. The selection of image color space and the use of facial alignment as preprocessing steps may collectively pose a significant impact on the accuracy and computational cost of facial emotion recognition, which is crucial to optimize the speed-accuracy trade-off. This paper proposed a deep learning-based facial emotion recognition pipeline that can be used to predict the emotion of detected face regions in video sequences. Five well-known state-of-the-art convolutional neural network architectures are used for training the emotion classifier to identify the network architecture which gives the best speed-accuracy trade-off. Two distinct facial emotion training datasets are prepared to investigate the effect of image color space and facial alignment on the performance of facial emotion recognition. Experimental results show that training a facial expression recognition model with grayscale-aligned facial images is preferable as it offers better recognition rates with lower detection latency. The lightweight MobileNet_v1 is identified as the best-performing model with WM=0.75 and RM=160 as its hyperparameters, achieving an overall accuracy of 86.42% on the testing video dataset.
Deep learning based masked face recognition in the era of the COVID-19 pandemicIJECEIAES
During the coronavirus disease 2019 (COVID-19) pandemic, monitoring for wearing masks obtains a crucial attention due to the effect of wearing masks to prevent the spread of coronavirus. This work introduces two deep learning models, the former based on pre-trained convolutional neural network (CNN) which called MobileNetv2, and the latter is a new CNN architecture. These two models have been used to detect masked face with three classes (correct, not correct, and no mask). The experiments conducted on benchmark dataset which is face mask detection dataset from Kaggle. Moreover, the comparison between two models is driven to evaluate the results of these two proposed models.
COMPRESSION BASED FACE RECOGNITION USING DWT AND SVMsipij
The biometric is used to identify a person effectively and employ in almost all applications of day to day
activities. In this paper, we propose compression based face recognition using Discrete Wavelet Transform
(DWT) and Support Vector Machine (SVM). The novel concept of converting many images of single person
into one image using averaging technique is introduced to reduce execution time and memory. The DWT is
applied on averaged face image to obtain approximation (LL) and detailed bands. The LL band coefficients
are given as input to SVM to obtain Support vectors (SV’s). The LL coefficients of DWT and SV’s are fused
based on arithmetic addition to extract final features. The Euclidean Distance (ED) is used to compare test
image features with database image features to compute performance parameters. It is observed that, the
proposed algorithm is better in terms of performance compared to existing algorithms.
Enhanced Face Detection Based on Haar-Like and MB-LBP FeaturesDr. Amarjeet Singh
The effective real-time face detection framework
proposed by Viola and Jones gained much popularity due its
computational efficiency and its simplicity. A notable
variant replaces the original Haar-like features with MBLBP (Multi-Block Local Binary Pattern) which are defined
by the local binary pattern operator, both detector types are
integrated into the OpenCV library. However, each
descriptor and its evaluation method has its own set of
strengths and setbacks. In this paper, an enhanced two-layer
face detector composed of both Haar-like and MB-LBP
features is presented. Haar-like features are employed as a
coarse filter but with a new evaluation involving dual
threshold. The already established MB-LBPs are arranged
as the fine filter of the detector. The Gentle AdaBoost
learning algorithm is deployed for the training of the
proposed detector to reach the classification and
performance potential. Experiments show that in the early
stages of classification, Haar features with dual threshold
are more discriminative than MB-LBP and original Haarlike features with respect to number of features required
and computation. Benchmarking the proposed detector
demonstrate overall 12% higher detection rate at 17% false
alarm over using MB-LBP features singly while performing
with ×3 speedup.
In this paper, an attempt has been made to extract texture
features from facial images using an improved method of
Illumination Invariant Feature Descriptor. The proposed local
ternary Pattern based feature extractor viz., Steady Illumination
Local Ternary Pattern (SIcLTP) has been used to extract texture
features from Indian face database. The similarity matching
between two extracted feature sets has been obtained using Zero
Mean Sum of Squared Differences (ZSSD). The RGB facial images
are first converted into the YIQ colour space to reduce the
redundancy of the RGB images. The result obtained has been
analysed using Receiver Operating Characteristic curve, and is
found to be promising. Finally the results are validated with
standard local binary pattern (LBP) extractor.
FACE EXPRESSION RECOGNITION USING CONVOLUTION NEURAL NETWORK (CNN) MODELS ijgca
This paper proposes the design of a Facial Expression Recognition (FER) system based on deep
convolutional neural network by using three model. In this work, a simple solution for facial expression
recognition that uses a combination of algorithms for face detection, feature extraction and classification
is discussed. The proposed method uses CNN models with SVM classifier and evaluates them, these models
are Alex-net model, VGG-16 model and Res-Net model. Experiments are carried out on the Extended
Cohn-Kanada (CK+) datasets to determine the recognition accuracy for the proposed FER system. In this
study the accuracy of AlexNet model compared with Vgg16 model and ResNet model. The result show that
AlexNet model achieved the best accuracy (88.2%) compared to other models.
Possibility fuzzy c means clustering for expression invariant face recognitionIJCI JOURNAL
Face being the most natural method of identification for humans is one of the most significant biometric
modalities and various methods to achieve efficient face recognition have been proposed. However the
changes in face owing to different expressions, pose, makeup, illumination, age bring about marked
variations in the facial image. These changes will inevitably occur and they can be controlled only till a
certain degree beyond which they are bound to happen and will affect the face thereby adversely impacting
the performance of any face recognition system. This paper proposes a strategy to improve the
classification methodology in face recognition by using Possibility Fuzzy C-Means Clustering (PFCM).
This clustering technique was used for face recognition due to its properties like outlier insensitivity which
make it a suitable candidate for use in designing such robust applications.PFCM is a hybridization of
Possibilistic C-Means (PCM) and Fuzzy C-Means (FCM) clustering algorithms. PFCM is a robust
clustering technique and is especially significant for its noise insensitivity. It has also resolved the
coincident clusters problem which is faced by other clustering techniques. Therefore the technique can also
be used to increase the overall robustness of a face recognition system and thereby increase its invariance
and make it a reliably usable biometric modality.
Content-based image retrieval based on corel dataset using deep learningIAESIJAI
A popular technique for retrieving images from huge and unlabeled image databases are content-based-image-retrieval (CBIR). However, the traditional information retrieval techniques do not satisfy users in terms of time consumption and accuracy. Additionally, the number of images accessible to users are growing due to web development and transmission networks. As the result, huge digital image creation occurs in many places. Therefore, quick access to these huge image databases and retrieving images like a query image from these huge image collections provides significant challenges and the need for an effective technique. Feature extraction and similarity measurement are important for the performance of a CBIR technique. This work proposes a simple but efficient deep-learning framework based on convolutional-neural networks (CNN) for the feature extraction phase in CBIR. The proposed CNN aims to reduce the semantic gap between low-level and high-level features. The similarity measurements are used to compute the distance between the query and database image features. When retrieving the first 10 pictures, an experiment on the Corel-1K dataset showed that the average precision was 0.88 with Euclidean distance, which was a big step up from the state-of-the-art approaches.
Facial image retrieval on semantic features using adaptive mean genetic algor...TELKOMNIKA JOURNAL
The emergence of larger databases has made image retrieval techniques an essential component and has led to the development of more efficient image retrieval systems. Retrieval can either be content or text-based. In this paper, the focus is on the content-based image retrieval from the FGNET database. Input query images are subjected to several processing techniques in the database before computing the squared Euclidean distance (SED) between them. The images with the shortest Euclidean distance are considered as a match and are retrieved. The processing techniques involve the application of the median modified Weiner filter (MMWF), extraction of the low-level features using histogram-oriented gradients (HOG), discrete wavelet transform (DWT), GIST, and Local tetra pattern (LTrP). Finally, the features are selected using Adaptive Mean Genetic Algorithm (AMGA). In this study, the average PSNR value obtained after applying the Wiener filter was 45.29. The performance of the AMGA was evaluated based on its precision, F-measure, and recall, and the obtained average values were respectively 0.75, 0.692, and 0.66. The performance matrix of the AMGA was compared to those of particle swarm optimization algorithm (PSO) and genetic algorithm (GA) and found to perform better; thus, proving its efficiency.
Depth-DensePose: an efficient densely connected deep learning model for came...IJECEIAES
Camera/image-based localization is important for many emerging applications such as augmented reality (AR), mixed reality, robotics, and self-driving. Camera localization is the problem of estimating both camera position and orientation with respect to an object. Use cases for camera localization depend on two key factors: accuracy and speed (latency). Therefore, this paper proposes Depth-DensePose, an efficient deep learning model for 6-degrees-of-freedom (6-DoF) camera-based localization. The Depth-DensePose utilizes the advantages of both DenseNets and adapted depthwise separable convolution (DS-Conv) to build a deeper and more efficient network. The proposed model consists of iterative depth-dense blocks. Each depth dense block contains two adapted DS-Conv with two kernel sizes 3 and 5, which are useful to retain both low-level as well as high-level features. We evaluate the proposed Depth-DensePose on the Cambridge Landmarks dataset, which shows that the Depth-DensePose outperforms the performance of related deep learning models for camera based localization. Furthermore, extensive experiments were conducted which proven the adapted DS-Conv is more efficient than the standard convolution. Especially, in terms of memory and processing time which is important to real-time and mobile applications.
The International Journal of Engineering and Science (IJES)theijes
The International Journal of Engineering & Science is aimed at providing a platform for researchers, engineers, scientists, or educators to publish their original research results, to exchange new ideas, to disseminate information in innovative designs, engineering experiences and technological skills. It is also the Journal's objective to promote engineering and technology education. All papers submitted to the Journal will be blind peer-reviewed. Only original articles will be published.
The face recognition technique gave in this study utilizes a reconfigurable organization of paramount threshold logic cells and can be utilized in the optional layer of a pixel exhibit. The technology used today for face recognition is neither either new nor particularly ancient. Face recognition is primarily employed for security reasons. Real-time applications have seen rapid growth in the demanding and fascinating field of face recognition. In recent years, face recognition has been the subject of intensive research. This report presents an up-to-date review of key human facial recognition studies. We begin by providing a general introduction of face recognition and its uses. The most recent facial recognition methods are then reviewed in the literature.
Similar to Robust face recognition using convolutional neural networks combined with Krawtchouk moments (20)
Bibliometric analysis highlighting the role of women in addressing climate ch...IJECEIAES
Fossil fuel consumption increased quickly, contributing to climate change
that is evident in unusual flooding and draughts, and global warming. Over
the past ten years, women's involvement in society has grown dramatically,
and they succeeded in playing a noticeable role in reducing climate change.
A bibliometric analysis of data from the last ten years has been carried out to
examine the role of women in addressing the climate change. The analysis's
findings discussed the relevant to the sustainable development goals (SDGs),
particularly SDG 7 and SDG 13. The results considered contributions made
by women in the various sectors while taking geographic dispersion into
account. The bibliometric analysis delves into topics including women's
leadership in environmental groups, their involvement in policymaking, their
contributions to sustainable development projects, and the influence of
gender diversity on attempts to mitigate climate change. This study's results
highlight how women have influenced policies and actions related to climate
change, point out areas of research deficiency and recommendations on how
to increase role of the women in addressing the climate change and
achieving sustainability. To achieve more successful results, this initiative
aims to highlight the significance of gender equality and encourage
inclusivity in climate change decision-making processes.
Voltage and frequency control of microgrid in presence of micro-turbine inter...IJECEIAES
The active and reactive load changes have a significant impact on voltage
and frequency. In this paper, in order to stabilize the microgrid (MG) against
load variations in islanding mode, the active and reactive power of all
distributed generators (DGs), including energy storage (battery), diesel
generator, and micro-turbine, are controlled. The micro-turbine generator is
connected to MG through a three-phase to three-phase matrix converter, and
the droop control method is applied for controlling the voltage and
frequency of MG. In addition, a method is introduced for voltage and
frequency control of micro-turbines in the transition state from gridconnected mode to islanding mode. A novel switching strategy of the matrix
converter is used for converting the high-frequency output voltage of the
micro-turbine to the grid-side frequency of the utility system. Moreover,
using the switching strategy, the low-order harmonics in the output current
and voltage are not produced, and consequently, the size of the output filter
would be reduced. In fact, the suggested control strategy is load-independent
and has no frequency conversion restrictions. The proposed approach for
voltage and frequency regulation demonstrates exceptional performance and
favorable response across various load alteration scenarios. The suggested
strategy is examined in several scenarios in the MG test systems, and the
simulation results are addressed.
Enhancing battery system identification: nonlinear autoregressive modeling fo...IJECEIAES
Precisely characterizing Li-ion batteries is essential for optimizing their
performance, enhancing safety, and prolonging their lifespan across various
applications, such as electric vehicles and renewable energy systems. This
article introduces an innovative nonlinear methodology for system
identification of a Li-ion battery, employing a nonlinear autoregressive with
exogenous inputs (NARX) model. The proposed approach integrates the
benefits of nonlinear modeling with the adaptability of the NARX structure,
facilitating a more comprehensive representation of the intricate
electrochemical processes within the battery. Experimental data collected
from a Li-ion battery operating under diverse scenarios are employed to
validate the effectiveness of the proposed methodology. The identified
NARX model exhibits superior accuracy in predicting the battery's behavior
compared to traditional linear models. This study underscores the
importance of accounting for nonlinearities in battery modeling, providing
insights into the intricate relationships between state-of-charge, voltage, and
current under dynamic conditions.
Smart grid deployment: from a bibliometric analysis to a surveyIJECEIAES
Smart grids are one of the last decades' innovations in electrical energy.
They bring relevant advantages compared to the traditional grid and
significant interest from the research community. Assessing the field's
evolution is essential to propose guidelines for facing new and future smart
grid challenges. In addition, knowing the main technologies involved in the
deployment of smart grids (SGs) is important to highlight possible
shortcomings that can be mitigated by developing new tools. This paper
contributes to the research trends mentioned above by focusing on two
objectives. First, a bibliometric analysis is presented to give an overview of
the current research level about smart grid deployment. Second, a survey of
the main technological approaches used for smart grid implementation and
their contributions are highlighted. To that effect, we searched the Web of
Science (WoS), and the Scopus databases. We obtained 5,663 documents
from WoS and 7,215 from Scopus on smart grid implementation or
deployment. With the extraction limitation in the Scopus database, 5,872 of
the 7,215 documents were extracted using a multi-step process. These two
datasets have been analyzed using a bibliometric tool called bibliometrix.
The main outputs are presented with some recommendations for future
research.
Use of analytical hierarchy process for selecting and prioritizing islanding ...IJECEIAES
One of the problems that are associated to power systems is islanding
condition, which must be rapidly and properly detected to prevent any
negative consequences on the system's protection, stability, and security.
This paper offers a thorough overview of several islanding detection
strategies, which are divided into two categories: classic approaches,
including local and remote approaches, and modern techniques, including
techniques based on signal processing and computational intelligence.
Additionally, each approach is compared and assessed based on several
factors, including implementation costs, non-detected zones, declining
power quality, and response times using the analytical hierarchy process
(AHP). The multi-criteria decision-making analysis shows that the overall
weight of passive methods (24.7%), active methods (7.8%), hybrid methods
(5.6%), remote methods (14.5%), signal processing-based methods (26.6%),
and computational intelligent-based methods (20.8%) based on the
comparison of all criteria together. Thus, it can be seen from the total weight
that hybrid approaches are the least suitable to be chosen, while signal
processing-based methods are the most appropriate islanding detection
method to be selected and implemented in power system with respect to the
aforementioned factors. Using Expert Choice software, the proposed
hierarchy model is studied and examined.
Enhancing of single-stage grid-connected photovoltaic system using fuzzy logi...IJECEIAES
The power generated by photovoltaic (PV) systems is influenced by
environmental factors. This variability hampers the control and utilization of
solar cells' peak output. In this study, a single-stage grid-connected PV
system is designed to enhance power quality. Our approach employs fuzzy
logic in the direct power control (DPC) of a three-phase voltage source
inverter (VSI), enabling seamless integration of the PV connected to the
grid. Additionally, a fuzzy logic-based maximum power point tracking
(MPPT) controller is adopted, which outperforms traditional methods like
incremental conductance (INC) in enhancing solar cell efficiency and
minimizing the response time. Moreover, the inverter's real-time active and
reactive power is directly managed to achieve a unity power factor (UPF).
The system's performance is assessed through MATLAB/Simulink
implementation, showing marked improvement over conventional methods,
particularly in steady-state and varying weather conditions. For solar
irradiances of 500 and 1,000 W/m2
, the results show that the proposed
method reduces the total harmonic distortion (THD) of the injected current
to the grid by approximately 46% and 38% compared to conventional
methods, respectively. Furthermore, we compare the simulation results with
IEEE standards to evaluate the system's grid compatibility.
Enhancing photovoltaic system maximum power point tracking with fuzzy logic-b...IJECEIAES
Photovoltaic systems have emerged as a promising energy resource that
caters to the future needs of society, owing to their renewable, inexhaustible,
and cost-free nature. The power output of these systems relies on solar cell
radiation and temperature. In order to mitigate the dependence on
atmospheric conditions and enhance power tracking, a conventional
approach has been improved by integrating various methods. To optimize
the generation of electricity from solar systems, the maximum power point
tracking (MPPT) technique is employed. To overcome limitations such as
steady-state voltage oscillations and improve transient response, two
traditional MPPT methods, namely fuzzy logic controller (FLC) and perturb
and observe (P&O), have been modified. This research paper aims to
simulate and validate the step size of the proposed modified P&O and FLC
techniques within the MPPT algorithm using MATLAB/Simulink for
efficient power tracking in photovoltaic systems.
Adaptive synchronous sliding control for a robot manipulator based on neural ...IJECEIAES
Robot manipulators have become important equipment in production lines, medical fields, and transportation. Improving the quality of trajectory tracking for
robot hands is always an attractive topic in the research community. This is a
challenging problem because robot manipulators are complex nonlinear systems
and are often subject to fluctuations in loads and external disturbances. This
article proposes an adaptive synchronous sliding control scheme to improve trajectory tracking performance for a robot manipulator. The proposed controller
ensures that the positions of the joints track the desired trajectory, synchronize
the errors, and significantly reduces chattering. First, the synchronous tracking
errors and synchronous sliding surfaces are presented. Second, the synchronous
tracking error dynamics are determined. Third, a robust adaptive control law is
designed,the unknown components of the model are estimated online by the neural network, and the parameters of the switching elements are selected by fuzzy
logic. The built algorithm ensures that the tracking and approximation errors
are ultimately uniformly bounded (UUB). Finally, the effectiveness of the constructed algorithm is demonstrated through simulation and experimental results.
Simulation and experimental results show that the proposed controller is effective with small synchronous tracking errors, and the chattering phenomenon is
significantly reduced.
Remote field-programmable gate array laboratory for signal acquisition and de...IJECEIAES
A remote laboratory utilizing field-programmable gate array (FPGA) technologies enhances students’ learning experience anywhere and anytime in embedded system design. Existing remote laboratories prioritize hardware access and visual feedback for observing board behavior after programming, neglecting comprehensive debugging tools to resolve errors that require internal signal acquisition. This paper proposes a novel remote embeddedsystem design approach targeting FPGA technologies that are fully interactive via a web-based platform. Our solution provides FPGA board access and debugging capabilities beyond the visual feedback provided by existing remote laboratories. We implemented a lab module that allows users to seamlessly incorporate into their FPGA design. The module minimizes hardware resource utilization while enabling the acquisition of a large number of data samples from the signal during the experiments by adaptively compressing the signal prior to data transmission. The results demonstrate an average compression ratio of 2.90 across three benchmark signals, indicating efficient signal acquisition and effective debugging and analysis. This method allows users to acquire more data samples than conventional methods. The proposed lab allows students to remotely test and debug their designs, bridging the gap between theory and practice in embedded system design.
Detecting and resolving feature envy through automated machine learning and m...IJECEIAES
Efficiently identifying and resolving code smells enhances software project quality. This paper presents a novel solution, utilizing automated machine learning (AutoML) techniques, to detect code smells and apply move method refactoring. By evaluating code metrics before and after refactoring, we assessed its impact on coupling, complexity, and cohesion. Key contributions of this research include a unique dataset for code smell classification and the development of models using AutoGluon for optimal performance. Furthermore, the study identifies the top 20 influential features in classifying feature envy, a well-known code smell, stemming from excessive reliance on external classes. We also explored how move method refactoring addresses feature envy, revealing reduced coupling and complexity, and improved cohesion, ultimately enhancing code quality. In summary, this research offers an empirical, data-driven approach, integrating AutoML and move method refactoring to optimize software project quality. Insights gained shed light on the benefits of refactoring on code quality and the significance of specific features in detecting feature envy. Future research can expand to explore additional refactoring techniques and a broader range of code metrics, advancing software engineering practices and standards.
Smart monitoring technique for solar cell systems using internet of things ba...IJECEIAES
Rapidly and remotely monitoring and receiving the solar cell systems status parameters, solar irradiance, temperature, and humidity, are critical issues in enhancement their efficiency. Hence, in the present article an improved smart prototype of internet of things (IoT) technique based on embedded system through NodeMCU ESP8266 (ESP-12E) was carried out experimentally. Three different regions at Egypt; Luxor, Cairo, and El-Beheira cities were chosen to study their solar irradiance profile, temperature, and humidity by the proposed IoT system. The monitoring data of solar irradiance, temperature, and humidity were live visualized directly by Ubidots through hypertext transfer protocol (HTTP) protocol. The measured solar power radiation in Luxor, Cairo, and El-Beheira ranged between 216-1000, 245-958, and 187-692 W/m 2 respectively during the solar day. The accuracy and rapidity of obtaining monitoring results using the proposed IoT system made it a strong candidate for application in monitoring solar cell systems. On the other hand, the obtained solar power radiation results of the three considered regions strongly candidate Luxor and Cairo as suitable places to build up a solar cells system station rather than El-Beheira.
An efficient security framework for intrusion detection and prevention in int...IJECEIAES
Over the past few years, the internet of things (IoT) has advanced to connect billions of smart devices to improve quality of life. However, anomalies or malicious intrusions pose several security loopholes, leading to performance degradation and threat to data security in IoT operations. Thereby, IoT security systems must keep an eye on and restrict unwanted events from occurring in the IoT network. Recently, various technical solutions based on machine learning (ML) models have been derived towards identifying and restricting unwanted events in IoT. However, most ML-based approaches are prone to miss-classification due to inappropriate feature selection. Additionally, most ML approaches applied to intrusion detection and prevention consider supervised learning, which requires a large amount of labeled data to be trained. Consequently, such complex datasets are impossible to source in a large network like IoT. To address this problem, this proposed study introduces an efficient learning mechanism to strengthen the IoT security aspects. The proposed algorithm incorporates supervised and unsupervised approaches to improve the learning models for intrusion detection and mitigation. Compared with the related works, the experimental outcome shows that the model performs well in a benchmark dataset. It accomplishes an improved detection accuracy of approximately 99.21%.
Developing a smart system for infant incubators using the internet of things ...IJECEIAES
This research is developing an incubator system that integrates the internet of things and artificial intelligence to improve care for premature babies. The system workflow starts with sensors that collect data from the incubator. Then, the data is sent in real-time to the internet of things (IoT) broker eclipse mosquito using the message queue telemetry transport (MQTT) protocol version 5.0. After that, the data is stored in a database for analysis using the long short-term memory network (LSTM) method and displayed in a web application using an application programming interface (API) service. Furthermore, the experimental results produce as many as 2,880 rows of data stored in the database. The correlation coefficient between the target attribute and other attributes ranges from 0.23 to 0.48. Next, several experiments were conducted to evaluate the model-predicted value on the test data. The best results are obtained using a two-layer LSTM configuration model, each with 60 neurons and a lookback setting 6. This model produces an R 2 value of 0.934, with a root mean square error (RMSE) value of 0.015 and a mean absolute error (MAE) of 0.008. In addition, the R 2 value was also evaluated for each attribute used as input, with a result of values between 0.590 and 0.845.
A review on internet of things-based stingless bee's honey production with im...IJECEIAES
Honey is produced exclusively by honeybees and stingless bees which both are well adapted to tropical and subtropical regions such as Malaysia. Stingless bees are known for producing small amounts of honey and are known for having a unique flavor profile. Problem identified that many stingless bees collapsed due to weather, temperature and environment. It is critical to understand the relationship between the production of stingless bee honey and environmental conditions to improve honey production. Thus, this paper presents a review on stingless bee's honey production and prediction modeling. About 54 previous research has been analyzed and compared in identifying the research gaps. A framework on modeling the prediction of stingless bee honey is derived. The result presents the comparison and analysis on the internet of things (IoT) monitoring systems, honey production estimation, convolution neural networks (CNNs), and automatic identification methods on bee species. It is identified based on image detection method the top best three efficiency presents CNN is at 98.67%, densely connected convolutional networks with YOLO v3 is 97.7%, and DenseNet201 convolutional networks 99.81%. This study is significant to assist the researcher in developing a model for predicting stingless honey produced by bee's output, which is important for a stable economy and food security.
A trust based secure access control using authentication mechanism for intero...IJECEIAES
The internet of things (IoT) is a revolutionary innovation in many aspects of our society including interactions, financial activity, and global security such as the military and battlefield internet. Due to the limited energy and processing capacity of network devices, security, energy consumption, compatibility, and device heterogeneity are the long-term IoT problems. As a result, energy and security are critical for data transmission across edge and IoT networks. Existing IoT interoperability techniques need more computation time, have unreliable authentication mechanisms that break easily, lose data easily, and have low confidentiality. In this paper, a key agreement protocol-based authentication mechanism for IoT devices is offered as a solution to this issue. This system makes use of information exchange, which must be secured to prevent access by unauthorized users. Using a compact contiki/cooja simulator, the performance and design of the suggested framework are validated. The simulation findings are evaluated based on detection of malicious nodes after 60 minutes of simulation. The suggested trust method, which is based on privacy access control, reduced packet loss ratio to 0.32%, consumed 0.39% power, and had the greatest average residual energy of 0.99 mJoules at 10 nodes.
Fuzzy linear programming with the intuitionistic polygonal fuzzy numbersIJECEIAES
In real world applications, data are subject to ambiguity due to several factors; fuzzy sets and fuzzy numbers propose a great tool to model such ambiguity. In case of hesitation, the complement of a membership value in fuzzy numbers can be different from the non-membership value, in which case we can model using intuitionistic fuzzy numbers as they provide flexibility by defining both a membership and a non-membership functions. In this article, we consider the intuitionistic fuzzy linear programming problem with intuitionistic polygonal fuzzy numbers, which is a generalization of the previous polygonal fuzzy numbers found in the literature. We present a modification of the simplex method that can be used to solve any general intuitionistic fuzzy linear programming problem after approximating the problem by an intuitionistic polygonal fuzzy number with n edges. This method is given in a simple tableau formulation, and then applied on numerical examples for clarity.
The performance of artificial intelligence in prostate magnetic resonance im...IJECEIAES
Prostate cancer is the predominant form of cancer observed in men worldwide. The application of magnetic resonance imaging (MRI) as a guidance tool for conducting biopsies has been established as a reliable and well-established approach in the diagnosis of prostate cancer. The diagnostic performance of MRI-guided prostate cancer diagnosis exhibits significant heterogeneity due to the intricate and multi-step nature of the diagnostic pathway. The development of artificial intelligence (AI) models, specifically through the utilization of machine learning techniques such as deep learning, is assuming an increasingly significant role in the field of radiology. In the realm of prostate MRI, a considerable body of literature has been dedicated to the development of various AI algorithms. These algorithms have been specifically designed for tasks such as prostate segmentation, lesion identification, and classification. The overarching objective of these endeavors is to enhance diagnostic performance and foster greater agreement among different observers within MRI scans for the prostate. This review article aims to provide a concise overview of the application of AI in the field of radiology, with a specific focus on its utilization in prostate MRI.
Seizure stage detection of epileptic seizure using convolutional neural networksIJECEIAES
According to the World Health Organization (WHO), seventy million individuals worldwide suffer from epilepsy, a neurological disorder. While electroencephalography (EEG) is crucial for diagnosing epilepsy and monitoring the brain activity of epilepsy patients, it requires a specialist to examine all EEG recordings to find epileptic behavior. This procedure needs an experienced doctor, and a precise epilepsy diagnosis is crucial for appropriate treatment. To identify epileptic seizures, this study employed a convolutional neural network (CNN) based on raw scalp EEG signals to discriminate between preictal, ictal, postictal, and interictal segments. The possibility of these characteristics is explored by examining how well timedomain signals work in the detection of epileptic signals using intracranial Freiburg Hospital (FH), scalp Children's Hospital Boston-Massachusetts Institute of Technology (CHB-MIT) databases, and Temple University Hospital (TUH) EEG. To test the viability of this approach, two types of experiments were carried out. Firstly, binary class classification (preictal, ictal, postictal each versus interictal) and four-class classification (interictal versus preictal versus ictal versus postictal). The average accuracy for stage detection using CHB-MIT database was 84.4%, while the Freiburg database's time-domain signals had an accuracy of 79.7% and the highest accuracy of 94.02% for classification in the TUH EEG database when comparing interictal stage to preictal stage.
Analysis of driving style using self-organizing maps to analyze driver behaviorIJECEIAES
Modern life is strongly associated with the use of cars, but the increase in acceleration speeds and their maneuverability leads to a dangerous driving style for some drivers. In these conditions, the development of a method that allows you to track the behavior of the driver is relevant. The article provides an overview of existing methods and models for assessing the functioning of motor vehicles and driver behavior. Based on this, a combined algorithm for recognizing driving style is proposed. To do this, a set of input data was formed, including 20 descriptive features: About the environment, the driver's behavior and the characteristics of the functioning of the car, collected using OBD II. The generated data set is sent to the Kohonen network, where clustering is performed according to driving style and degree of danger. Getting the driving characteristics into a particular cluster allows you to switch to the private indicators of an individual driver and considering individual driving characteristics. The application of the method allows you to identify potentially dangerous driving styles that can prevent accidents.
Hyperspectral object classification using hybrid spectral-spatial fusion and ...IJECEIAES
Because of its spectral-spatial and temporal resolution of greater areas, hyperspectral imaging (HSI) has found widespread application in the field of object classification. The HSI is typically used to accurately determine an object's physical characteristics as well as to locate related objects with appropriate spectral fingerprints. As a result, the HSI has been extensively applied to object identification in several fields, including surveillance, agricultural monitoring, environmental research, and precision agriculture. However, because of their enormous size, objects require a lot of time to classify; for this reason, both spectral and spatial feature fusion have been completed. The existing classification strategy leads to increased misclassification, and the feature fusion method is unable to preserve semantic object inherent features; This study addresses the research difficulties by introducing a hybrid spectral-spatial fusion (HSSF) technique to minimize feature size while maintaining object intrinsic qualities; Lastly, a soft-margins kernel is proposed for multi-layer deep support vector machine (MLDSVM) to reduce misclassification. The standard Indian pines dataset is used for the experiment, and the outcome demonstrates that the HSSF-MLDSVM model performs substantially better in terms of accuracy and Kappa coefficient.
About
Indigenized remote control interface card suitable for MAFI system CCR equipment. Compatible for IDM8000 CCR. Backplane mounted serial and TCP/Ethernet communication module for CCR remote access. IDM 8000 CCR remote control on serial and TCP protocol.
• Remote control: Parallel or serial interface.
• Compatible with MAFI CCR system.
• Compatible with IDM8000 CCR.
• Compatible with Backplane mount serial communication.
• Compatible with commercial and Defence aviation CCR system.
• Remote control system for accessing CCR and allied system over serial or TCP.
• Indigenized local Support/presence in India.
• Easy in configuration using DIP switches.
Technical Specifications
Indigenized remote control interface card suitable for MAFI system CCR equipment. Compatible for IDM8000 CCR. Backplane mounted serial and TCP/Ethernet communication module for CCR remote access. IDM 8000 CCR remote control on serial and TCP protocol.
Key Features
Indigenized remote control interface card suitable for MAFI system CCR equipment. Compatible for IDM8000 CCR. Backplane mounted serial and TCP/Ethernet communication module for CCR remote access. IDM 8000 CCR remote control on serial and TCP protocol.
• Remote control: Parallel or serial interface
• Compatible with MAFI CCR system
• Copatiable with IDM8000 CCR
• Compatible with Backplane mount serial communication.
• Compatible with commercial and Defence aviation CCR system.
• Remote control system for accessing CCR and allied system over serial or TCP.
• Indigenized local Support/presence in India.
Application
• Remote control: Parallel or serial interface.
• Compatible with MAFI CCR system.
• Compatible with IDM8000 CCR.
• Compatible with Backplane mount serial communication.
• Compatible with commercial and Defence aviation CCR system.
• Remote control system for accessing CCR and allied system over serial or TCP.
• Indigenized local Support/presence in India.
• Easy in configuration using DIP switches.
HEAP SORT ILLUSTRATED WITH HEAPIFY, BUILD HEAP FOR DYNAMIC ARRAYS.
Heap sort is a comparison-based sorting technique based on Binary Heap data structure. It is similar to the selection sort where we first find the minimum element and place the minimum element at the beginning. Repeat the same process for the remaining elements.
Industrial Training at Shahjalal Fertilizer Company Limited (SFCL)MdTanvirMahtab2
This presentation is about the working procedure of Shahjalal Fertilizer Company Limited (SFCL). A Govt. owned Company of Bangladesh Chemical Industries Corporation under Ministry of Industries.
Cosmetic shop management system project report.pdfKamal Acharya
Buying new cosmetic products is difficult. It can even be scary for those who have sensitive skin and are prone to skin trouble. The information needed to alleviate this problem is on the back of each product, but it's thought to interpret those ingredient lists unless you have a background in chemistry.
Instead of buying and hoping for the best, we can use data science to help us predict which products may be good fits for us. It includes various function programs to do the above mentioned tasks.
Data file handling has been effectively used in the program.
The automated cosmetic shop management system should deal with the automation of general workflow and administration process of the shop. The main processes of the system focus on customer's request where the system is able to search the most appropriate products and deliver it to the customers. It should help the employees to quickly identify the list of cosmetic product that have reached the minimum quantity and also keep a track of expired date for each cosmetic product. It should help the employees to find the rack number in which the product is placed.It is also Faster and more efficient way.
Welcome to WIPAC Monthly the magazine brought to you by the LinkedIn Group Water Industry Process Automation & Control.
In this month's edition, along with this month's industry news to celebrate the 13 years since the group was created we have articles including
A case study of the used of Advanced Process Control at the Wastewater Treatment works at Lleida in Spain
A look back on an article on smart wastewater networks in order to see how the industry has measured up in the interim around the adoption of Digital Transformation in the Water Industry.
Robust face recognition using convolutional neural networks combined with Krawtchouk moments
1. International Journal of Electrical and Computer Engineering (IJECE)
Vol. 13, No. 4, August 2023, pp. 4052~4067
ISSN: 2088-8708, DOI: 10.11591/ijece.v13i4.pp4052-4067 4052
Journal homepage: http://ijece.iaescore.com
Robust face recognition using convolutional neural networks
combined with Krawtchouk moments
Yassir El Madmoune, Ilham El Ouariachi, Khalid Zenkouar, Azeddine Zahi
Laboratory of Intelligent Systems and Application, Faculty of Sciences and Technology, University Sidi Mohamed Ben Abdellah, Fez,
Morocco
Article Info ABSTRACT
Article history:
Received Jul 16, 2022
Revised Oct 25, 2022
Accepted Nov 6, 2022
Face recognition is a challenging task due to the complexity of pose
variations, occlusion and the variety of face expressions performed by distinct
subjects. Thus, many features have been proposed, however each feature has
its own drawbacks. Therefore, in this paper, we propose a robust model called
Krawtchouk moments convolutional neural networks (KMCNN) for face
recognition. Our model is divided into two main steps. Firstly, we use 2D
discrete orthogonal Krawtchouk moments to represent features. Then, we fed
it into convolutional neural networks (CNN) for classification. The main goal
of the proposed approach is to improve the classification accuracy of noisy
grayscale face images. In fact, Krawtchouk moments are less sensitive to
noisy effects. Moreover, they can extract pertinent features from an image
using only low orders. To investigate the robustness of the proposed approach,
two types of noise (salt and pepper and speckle) are added to three datasets
(YaleB extended, our database of faces (ORL), and a subset of labeled faces
in the wild (LFW)). Experimental results show that KMCNN is flexible and
performs significantly better than using just CNN or when we combine it with
other discrete moments such as Tchebichef, Hahn, Racah moments in most
densities of noises.
Keywords:
Convolutional neural networks
Face recognition
Image moments
Krawtchouk moments
Noisy conditions
This is an open access article under the CC BY-SA license.
Corresponding Author:
Yassir El Madmoune
Laboratory of Intelligent Systems and Application, Faculty of Sciences and Technology, University Sidi
Mohamed Ben Abdellah
Fez, Morocco
Email: yassir.elmadmoune@usmba.ac.ma
1. INTRODUCTION
Face recognition is an essential aspect of biometric technologies [1], [2]. that has received significant
attention due to the fast development of technology such as digital cameras [3], the Internet [4], and mobile
devices [5], as well as the rising desire for security [6]–[9]. Face recognition offers various benefits over other
biometric systems, including natural, non-intrusive, and simple. However, face recognition has become one of
the most challenging pattern recognition problems, owing the wide range of lighting circumstances, face
expression, head size, pose variation, complex background, motion blurring, noisy conditions, and other
environmental variables that could reduce recognition performance [10].
Three primary categories can be used to categorize the various approaches that have been utilized for
face recognition: holistic approaches [11]–[21], feature-based approaches [22]–[30] and hybrid approaches
[31]–[34]. In early 1990, researchers in face recognition field started using holistic approaches, i.e., facial
detection systems use the entire face region as input to accomplish face recognition. In this approach, we find
two sub-categories of techniques: the first one is based on linear methods like Eigenfaces principal component
analysis (PCA) [11], [12], Fisherfaces linear discriminative analysis (LDA) [13], [14], independent component
2. Int J Elec & Comp Eng ISSN: 2088-8708
Robust face recognition using convolutional neural networks combined … (Yassir El Madmoune)
4053
analysis (ICA) [15], discrete wavelet transform (DWT) [16] and discrete cosine transform (DCT) [17]. The
second technique is based on non-linear methods such as Kernel PCA (KPCA) [18], kernel linear discriminant
analysis (KLDA) [19], Gabor-KLDA [20], and CNN [21]. In the first decade of the 21st century, studies have
focused on feature-based approaches, and could possibly be separated into two distinct types: local appearance-
based techniques that consider the facial image as a collection of discrete vectors with low dimensions and
focus on crucial parts of the face like the nose, mouth, and eyes to create additional information and make face
recognition easier. Local binary pattern (LBP) [22], histogram of oriented gradients (HOG) [23], correlation
filters (joint transform correlator (JTC) [24], VanderLugt correlator (VLC) [25]) and discrete orthogonal
moments (DOM) [26] are the most methods used in this sub category. In the second sub-category, keypoints-
based techniques are utilized to detect particular geometric characteristics based on the geometry of the facial
features (e.g., the space between the eyes, the circumference of the head) using algorithms like scale-invariant
feature transform (SIFT) [27] and descriptors like speeded-up robust features (SURF) [28], binary robust
independent elementary features (BRIEF) [29] and fast retina keypoint (FREAK) [30]. In early 2010, the face
recognition community focused on hybrid approaches that combine local and subspace features to maximize
the strengths of both types of approaches which could provide enhanced performance in face recognition
systems, such as Gabor wavelet and linear discriminant analysis (GW-LDA) [31], multi-sub-region-based
correlation filter bank (MS-CFB) [32], CNNs and stacked auto-encoder (SAE) [33], advanced correlation
filters and Walsh LBP (WLBP) [34], Figure 1 shows a brief organization of the previous mentioned approaches.
Recently, deep learning (DL) and more specifically convolution neural networks (CNN) is the most commonly
methodology used for extracting features in face recognition, it has significant advantages due to its learning
ability, generalization, and robustness [14], [15]. Deep and extensive neural networks have demonstrated
remarkable performance with massive training datasets and the computing capacity of graphical processing
units (GPUs); It could generate the fundamental feature representation of data and create high-level features
from the low-level pixels.
Ding et al. [35] presented the noise resistant network (NR-network), a deep learning network-based
system that extracts low-level and high-level face characteristics using a multi-input structure; they used a
downscaling approach to reduce the resolution of their dataset in order to accelerate the processing, focusing
on facial recognition in noisy conditions. However, basic design and massive pooling operations are lost certain
facial features. As a result, such a system will not be able to recognize faces in noisy environments.
Meng et al. [36] presented a deep CNN with sub-networks for denoising and recognizing faces under noise;
unlike traditional approaches, which train the two sub-networks separately, this method trains them together;
hence, it requires more time. Wu et al. [37] proposed a light CNN framework based on three networks that
reduce the computational costs and the number of parameters to train a 256-D compact embedding from
enormous face data with several noisy labels, Ma et al. [38] introduce a robust local binary pattern (LBP)
guiding pooling (G-RLBP) mechanism to enhance the accuracy of CNNs models while effectively reducing
the noise effects.
Dimensionality reduction and feature extraction are essential parts of any facial recognition system.
Despite the fact that face images have a high dimensionality despite their small size, which leads to a significant
amount of computational time, complexity, and memory occupation; the performance of any classifier is mainly
determined by the good discriminating features included inside the face image [39], [40]. In this sense, the
presence of noisy training data can harm the ultimate performance of trained convolutional neural networks.
Although a recent research demonstrated that deep CNNs work well even on noisy samples with sufficient clean
data [41], this conclusion is not always applicable in face recognition. Experimental tests indicate that noisy data
appears to reduce the performance of trained face recognition CNNs [42]. To overcome these constraints and
improve performance, another feature extraction technique that can deal with noise must be used.
Orthogonal moments are robust in the presence of image noise and have a near-zero redundancy
measure in a feature set. In this respect, 2D DOM that are based on the Krawtchouk polynomials [43] has the
ability to extract local features from any region of interest in an image in addition to the global feature extraction
capability. Apostolidis and Papakostas [44] showed that using Krawtchouk moments as an image local
descriptor and a watermarking attack can affects the accuracy of deep learning models when it applied in
medical images. Amakdouf et al. [45] came up with quaternion radial Krawtchouk moments that could be
useful in the field of color image analysis by showing a good representation capability and robustness to
different noises. Hassan et al. [46] demonstrated that invariant quaternion Krawtchouk moments are more
effective than continuous orthogonal moments at representing images and showed more stability against the
translation, rotation, and scaling transformation. Rahman et al. [47] introduced a new method for face
recognition in which sparse representation of face images is created by selecting a discriminatory set of
orthogonal Krawtchouk moments. Following the considerations presented above, the Krawtchouk DOM are
investigated for grayscale face image recognition. The essence of our suggested model is to employ Krawtchouk
moments as a fast and accurate object descriptor; the whole face shape moments could be computed and fed it as
input layer to a convolutional neural network, the robustness of our proposed model is tested on small and large
3. ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 13, No. 4, August 2023: 4052-4067
4054
size databases with the presence of two types of noise and compared with CNN combined with others 2D DOM
and without them. The main contributions of this study are summarized as follows:
- A new architecture named Krawtchouk moments convolutional neural networks (KMCNN), defined by
Krawtchouk orthogonal polynomials, is introduced for the first time in this paper.
- A robust face recognition approach against various types of noises is proposed.
- An application of the suggested KMCNN model for face reconstruction and recognition is presented.
The remainder of this paper is structure as follows. Section 2 a brief review of 2D Krawtchouk
orthogonal moments and the process of creating image moments. Section 3 describes the proposed KMCNN
model and its architecture. The databases are considered in section 4. Experiments and results details are also
conducted to evaluate the KMCNN compared with CNN only and its combination with other 2D orthogonal
moments in this section. Section 5 concludes this paper.
Figure 1. Summary of face recognition approaches
2. 2D KRAWTCHOUK MOMENTS
Krawtchouk moments are a set of orthogonal moments based on the discrete Krawtchouk polynomials
defined over the coordinate image space. Their implementation does not involve any numerical approximation.
In this section, we will give a brief formulation of 2D weighted Krawtchouk moments, including polynomials and
describe their capacity to capture significant features from images with a significant dimensionality reduction.
2.1. Krawtchouk polynomials
The Krawtchouk polynomials were initially presented by Krawtchouk [48], and recently utilized by
Yap et al. [49] image analysis fields. The orthogonality relation of the Krawtchouk discrete polynomials is
given by (1).
∑𝑁−1
𝑥=0 𝑤𝑘(𝑥; 𝑝, 𝑁)𝑘𝑛(𝑥; 𝑝, 𝑁)𝑘𝑚(𝑥; 𝑝, 𝑁) = 𝜌𝑘(𝑛; 𝑝, 𝑁)𝛿𝑛𝑚 𝑛, 𝑚 = 1, … , 𝑁, (1)
where 𝑤𝑘(𝑥; 𝑝, 𝑁) is the weighting function defined as (2):
𝑤𝑘(𝑥; 𝑝, 𝑁 − 1) = (
𝑁 − 1
𝑥
) 𝑝𝑥
(1 − 𝑝)𝑁−1−𝑥
, (2)
with the norm function is:
𝜌𝑘(𝑛; 𝑝, 𝑁 − 1) = (−1)𝑛
(
1−𝑝
𝑝
)
𝑛 𝑛!
(1−𝑁)𝑛
. (3)
4. Int J Elec & Comp Eng ISSN: 2088-8708
Robust face recognition using convolutional neural networks combined … (Yassir El Madmoune)
4055
Using the definition above, Yap et al. [49] presents the recurrent formula by using the normalized Krawtchouk
polynomials.
𝑘𝑛(𝑥; 𝑝, 𝑁 − 1) = 𝐴𝑛𝑘𝑛−1(𝑥; 𝑝, 𝑁 − 1) − 𝐵𝑛𝑘
̅𝑛−2(𝑥; 𝑝, 𝑁 − 1)
𝑘
̅0(𝑥; 𝑝, 𝑁 − 1) = 𝑤𝑘(𝑥; 𝑝, 𝑁 − 1)
𝑘
̅1(𝑥; 𝑝, 𝑁 − 1) = 𝑤𝑘(𝑥; 𝑝, 𝑁 − 1)
(𝑁−1)𝑝−𝑥
√(𝑁−1)𝑝(1−𝑝)
with 𝐴𝑛 =
((𝑁−1𝑝−2(𝑛−1)𝑝+𝑛−1−𝑥)
√𝑝(1−𝑝)𝑛(𝑁−𝑛)
and 𝐵𝑛 = √
(𝑛−1)(𝑁−𝑛+1)
(𝑁−𝑛)𝑛
(4)
Figures 2(a) and (b) show the weighted Krawtchouk polynomials up to the 7th
degree for p=0.5 and
p=0.2, respectively. The graphs illustrate the impact of the localization parameter p, which permits the
polynomials to be moved to the appropriate location.
(a) (b)
Figure 2. Weighted Krawtchouk polynomials up to the 7th
degree for N=168, (a) p=0.5 and (b) p=0.2
2.2. Krawtchouk moments
In general, moments are defined as scalar values, that are consistent and efficient data descriptors [50].
They may be used to represent 1D signals like voice and 2D signals such as images without information
redundancy in the moment set and to detect slight signal intensity variations [51]. For a two-dimensional signal
with intensity function f(x, y) of size N1×N2, Krawtchouk moments 𝜓𝑛𝑚can be defined as (5) [50], [52]:
𝜓𝑛𝑚 = & ∑𝑁1−1
𝑥=0 ∑𝑁2−1
𝑦=0 𝐾
̅𝑛(𝑥; 𝑝, 𝑁1 − 1)𝐾
̅𝑚(𝑦; 𝑝, 𝑁2 − 1)𝑓(𝑥, 𝑦)
𝑛 = &0,1, … , 𝑀1 − 1 and 𝑚 = 0,1, … , 𝑀2 − 1,
(5)
where M1 and M2 are the maximum moment orders used to describe the intensity signal f(x, y). To recover
f(x, y) from Krawtchouk Moments, the (6) is used:
𝑓
̂(𝑥, 𝑦) = ∑𝑀1−1
𝑛=0 ∑𝑀2−1
𝑚=0 𝐾
̅𝑛(𝑥; 𝑝, 𝑁1 − 1)𝐾
̅𝑚(𝑦; 𝑝, 𝑁2 − 1)𝜓𝑛𝑚
𝑥 = 0,1, … , 𝑁1 − 1 and 𝑦 = 0,1, … , 𝑁2 − 1,
(6)
where 𝑓
̂(𝑥, 𝑦) is the reconstructed function, 𝑓
̂(𝑥, 𝑦) = 𝑓(𝑥, 𝑦) when all moments are taken into account
throughout the reconstruction process.
Figures 3(a) to (c) shows respectively a sample of an original face image from YaleB database [53],
a noisy image when we mix the original image with 5% of salt and pepper and speckle noises. Figures 4 and 5
show respectively reconstructions of face images mixed with salt and pepper and speckle noises, where
subfigures (a) to (j) show the reconstruction up to orders 168 from 10 to 160 with 20 increments by using 2D
Krawtchouk moments and noisy images shown in Figures 3(b) and (c). We choose p = 0.5 to obtain the highest
representation; In the early stages, we can see a more striking resemblance between the noisy images and the
reconstructed ones. This indicates that 2D Krawtchouk moments have the ability to extract more information
from images in a smaller space, which means that instead of using the original picture, we may employ image
moments to reduce dimensionality and extract meaningful features for classification.
5. ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 13, No. 4, August 2023: 4052-4067
4056
(a) (b) (c)
Figure 3. sample of image with/without noise from YaleB database, (a) original image, (b) with 5% of salt
and pepper noise, and (c) with 5% of speckle-noise
(a) (b) (c) (d) (e)
(f) (g) (h) (i) (j)
Figure 4. Reconstruction results using 2D Krawtchouk moments and an image from YaleB dataset mixed
with 5% of salt and pepper noise, (a) to (j) from 10 to 160 with 20 increments
(a) (b) (c) (d) (e)
(f) (g) (h) (i) (j)
Figure 5. Reconstruction results using 2D Krawtchouk moments and an image from YaleB dataset mixed
with 5% of speckle noise, (a) to (j) from 10 to 160 with 20 increments
6. Int J Elec & Comp Eng ISSN: 2088-8708
Robust face recognition using convolutional neural networks combined … (Yassir El Madmoune)
4057
3. PROPOSED MODEL
In this work, we presented a novel architecture for face recognition problems named KMCNN that
combine the idea of orthogonal moments with the 2D CNN model, as shown in Figure 6. Indeed, duo to
Krawtchouk moments property for representing face images in lower orders without redundancy, as
demonstrated in the previous section; which facilitates the production of small 2D moments matrices that are
inserted into a 2D convolutional neural network. Therefore, we get two benefits from this combination, the
processing complexity is significantly reduced, and the computational speed is increased. Table 1 provides a
summary of the principal model layers, and the suggested architecture design is organized as follows:
- 2D Krawtchouk Moment layer: As the first layer of the KMCNN, Krawtchouk discrete orthogonal moments
compute the input image moments by using (5) and provide a matrix whose size is proportional to the
moment order value. This layer optimizes the image representation and decreases the processing dimension
significantly. The subsequent 2D convolutional layer is then given this matrix of moments.
- 2D Convolution layer: the purpose behind this layer is to recognize the presence of a set of features in the
moment matrix rather than the original image, by the use of 2D convolution operators. The output
activation value 𝑎(𝑖, 𝑗)𝐿
at position (i,j) is calculated by (7).
a(i, j)L
= f (∑i+N−1
x=i ∑j+N−1
y=j
∑S−1
s=0 Ws,x,yMs,x,y + bL
) (7)
where the matrix of moments M convolves with the 𝐿𝑡ℎ
filter with a size of N×N, S is the number of input
channels, W is the weight matrix with size (C, N, N), i,j are the indices of the output position, x,y are the indices
of the input position. 𝑓 is the activation function.
- Activation functions ReLU: The output feature maps from the convolution layer are given a non-linear
transformation when they are sent through the activation layer. By transforming the data into a non-linear
format, it facilitates the identification of complex features that cannot be explained using a linear
combination of a regression technique. The most regularly used non-linear functions are sigmoid and
hyperbolic tangent; in this work, the rectified linear unit (ReLU) defined by f(x)=max(0, x) is used, since it
improves the non-linearity, avoids network saturation and speeds up training time networks [54]–[56].
- Batch normalization: Allows each convolutional layer to learn more independently. This layer normalizes
the output of the preceding layers to enhance their learning process and prevent overfitting and divergence
risks [57].
- 2D Max-pooling layer: A pooling layer is typically added following the convolution layers, to decrease the
size of the feature maps. Consequently, the number of network parameters, as a result the computation time
is accelerated and the chance of the model falling in overfitting is diminished.
- Fully connected layer: Is the last layer in our proposed KMCNN, this layer performs a linear combination
on the data received from the preceding layers, and then applies the softmax function to produce the
probability of each class as a new output vector.
Figure 6. 2D KMCNN architecture
7. ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 13, No. 4, August 2023: 4052-4067
4058
Table 1. Specifics of the suggested model.
Layer Purpose Filter Number of filters Activation
0 Input image - - 𝑁𝑥𝑁
1 Moments layer - - 𝑛𝑥𝑛
2 Conv+ReLU 3x3 5 𝑛𝑥𝑛𝑥5
3 MaxPooling 2x2 - 𝑛/2𝑥𝑛/2 ∗ 5
4 Conv+ReLU 3x3 10 𝑛/2𝑥𝑛/2 ∗ 10
5 MaxPooling 2x2 - 𝑛/4𝑥𝑛/4 ∗ 10
6 Fully connected - - 1,000
7 Softmax - - number of subjects
4. EXPERIMENTS AND RESULTS
This section presents details about experiments conducted on face images using the proposed model
KMCNN and provides a thorough description of the databases. The experiments are divided into two parts, the
first one was conducted in free noise environment where the second was performed with presence of noise.
Additionally, this section discusses the recognition accuracies obtained.
4.1. Experiments
In this sub-section, we evaluated the classification performance of the proposed model by carrying
out a number of relevant experiments using YaleB extended database [53], our database of faces (ORL)
database [58] and a subset of 10 classes from labeled faces in the wild (LFW) database [10]. By randomly
dividing each database into 70% for training and 30% for testing, the efficacy of the suggested model is
examined, the results are properly compared to several 2D orthogonal moment-based approaches. All
experiments were conducted in the cloud using Google Collaboratory with 2.20 GHz, Intel(R) Xeon(R) CPU,
NVIDIA-SMI GPU and 13 GB of RAM. The evaluation of the recognition rate of the suggested model
with/without noise is structured around three primary comparisons:
- First, a comparison of the accuracy of Tchebichef, Krawtchouk, Hahn, and Racah moments as an input
layer of the suggested CNN architecture was conducted using YaleB extended, ORL, and a subset of LFW
database without any presence of noise. A comparison with existing methods is also presented.
- Second, we compared our suggested model KMCNN against CNN only, in noisy environments using two
forms of noise (salt and pepper and speckle)
- Third, we have compared our proposed model with CNN combined with other 2D discrete orthogonal
moments and Krawtchouk combined with pre-trained VGG16 model [59] in the same noisy environments.
In addition, we used different densities of noise to test our model in noisy environments, by varying
the salt and pepper noise densities from 1% to 5% and Speckle noise by varying the variance value from 0.1 to
0.5 and a fixed the mean at 0.
4.2. Datasets
In the course of those experiments, three face image databases are utilized, in order to investigate the
recognition rate performance. The two first databases contain grayscale images, whereas the third provide red,
green and blue (RGB) images that have been transformed to grayscale format. The selected databases are as
follows:
- The extended YaleB database: comprises 16,128 pictures of 28 individuals in 9 different positions and 64
lighting settings. This database follows the same data structure of the YaleB Database. In contrast to the
original YaleB database consisting of 10 participants, the extended database was originally revealed by Lee
et al. [53]. Since we are not concentrating on position variation, only the frontal face image of each subject
with 64 different illuminations will be selected, totaling 2,432 images from 38 distinct subjects. Manual
alignment, cropping, and resizing to 168 by 192 pixels is performed on every image used in the experiments.
Figure 7 depicts a selection of facial image instances.
- The ORL database [58] consists of 400 images in total, including 40 persons with 10 unique image
(4 females and 36 males), the images were captured at various periods, changing the illumination, face
gestures (open/closed eyes, smiling/not smiling), and facial characteristics (glasses/no glasses). Figure 8
shows that all of the images were taken with the people standing up and facing forward against a black
background. The dimensions of each image are 70 by 80 pixels, and there are 256 levels of gray for each
individual pixel.
- The LFW database [10] includes 13,233 face images gathered from the internet. This collection contains
5,749 identities for 1,680 individuals with two or more images. In this work we choose a subset of
10 classes of people that have the most available images with total of 1456 images and the dimensions of
each image are 240 by 240 pixels. Figure 9 shows an example of images used.
8. Int J Elec & Comp Eng ISSN: 2088-8708
Robust face recognition using convolutional neural networks combined … (Yassir El Madmoune)
4059
Figure 7. Examples of images from YaleB database
Figure 8. Examples of images from ORL database Figure 9. Examples of images from a subset of
LFW database
4.3. Results
4.3.1. Face recognition with free noise
Experiment 1: comparison between orthogonal moments.
As mentioned in the first sub-section, we begin our experiments by analyzing the classification
performance of the suggested CNN architecture using original images from YaleB [53], ORL [58], and LFW
[10], and compared the results with CNN combined with Tchebichef moments [60], Hahn moments [61], Racah
Moments [62], the corresponding classification accuracy results using the databases mentioned before started
from lower orders to the maximum order are listed and summarized in Tables 2 to 4, each column in the tables
represents the performance in terms of accuracy of the suggested CNN architecture combined with a different
type of 2D orthogonal moments.
Based on results from Table 2, the greatest score is achieved at the order (168,168) using Krawtchouk
moments as an input layer on YaleB database with 92.03% of accuracy, followed by Tchebichef moments with
a precision of 87.98% at the order (140,140), Hahn moments with 86.36% of accuracy at the order (160,160)
and Racah moments with 82.99%. Nevertheless, we can see that Krawtchouk moments give interesting results
starting from order 60 by surpassing 90% in terms of accuracy. However, As shown in Table 3, the fusion of
CNN and Krawtchouk moments does not surpass other fusions of CNN with 2D discrete moments when we
tested it on small-size face images from ORL database. Table 4 clearly shows that the suggested KMCNN
outperforms the rest of models based on other 2D discrete orthogonal moments; we can clearly notice that the
combination of Krawtchouk moments with convolutional neural networks gives 74.30% in terms of accuracy
at order 20 when we test it on original images of LFW database. As a conclusion achieved from Tables 2 to 4,
we can say that face image recognition by using CNN and Krawtchouk moments as input layer was
significantly improved compared with results obtained using CNN combined with Tchebichef, Racah and Hahn
moments.
Experiment 2: comparison with the state-of-the-art methods.
In order to illustrate the effectiveness of the suggested model, the classification accuracy is compared
with the state-of-the-art approaches for face recognition. Table 5 shows the comparative analysis of the
recognition rate, between the suggested KMCNN and the other approaches for the extended Yale B and the
ORL databases. Each row from the table shows the method and the corresponding accuracy achieved, whereas
the last row represents the accuracy of the KMCNN.
9. ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 13, No. 4, August 2023: 4052-4067
4060
Table 2. Classification accuracies for different orders using Tchebichef, Krawtchouk, Hahn and Racah
moments tested on YaleB database
Order Tchebichef moments Krawtchouk moments Hahn moments Racah moments
10 43.45 28.74 47.23 38.86
20 71.52 66.93 71.65 68.82
40 85.15 89.47 82.18 80.43
60 85.34 90.41 84.34 82.45
80 84.88 90.55 84.34 82.99
100 84.21 90.95 80.56 82.45
120 82.18 91.22 82.72 81.37
140 87.98 90.95 81.24 80.29
160 85.42 91.76 86.36 78.40
168 85.69 92.03 85.96 77.59
Table 3. Classification accuracies for different orders using Tchebichef, Krawtchouk, Hahn and Racah
moments tested on ORL database
Order Tchebichef moments Krawtchouk moments Hahn moments Racah moments
10 56.1 28.05 63.41 47.56
20 91.46 78.05 93.9 87.8
30 95.12 96.34 93.9 95.12
40 96.34 97.56 97.56 96.34
50 98.78 95.12 97.56 96.34
60 97.56 95.12 97.56 97.56
70 96.34 95.12 97.56 96.34
Table 4. Classification accuracies for different orders using Tchebichef, Krawtchouk, Hahn and Racah
moments tested on LFW database
Order Tchebichef moments Krawtchouk moments Hahn moments Racah moments
10 41.44 56.16 38.36 41.44
15 39.73 68.84 39.04 34.93
20 44.86 74.32 36.99 40.41
25 46.23 72.26 37.67 40.41
30 50.68 70.55 42.81 40.75
40 53.08 64.04 47.95 47.6
60 60.96 59.93 52.05 50.0
80 64.38 56.16 56.16 56.85
100 63.7 54.11 58.22 55.82
150 61.99 50.34 57.53 56.16
200 63.36 54.79 58.22 57.53
250 63.7 52.4 57.53 56.16
Table 5. the comparison between the state-of-the-art methods and the KMCNN for the YaleB and ORL
databases
YaleB ORL
Methods Accuracy Methods Accuracy
LSP [63] 85.6 PCA [18] 93.91
POEM [64] 90.5 2DHOG [65] 97
LBP [66] 78.6 SIFT [67] 97
GENet [68] 84.21 SURF [69] 88
Gabor [70] 87.19 HOG + ConvNet [71] 95.5
KMCNN 92.03 KMCNN 97.56
To validate the efficacy of our suggested approach, we compared it to other methods were used the
Extended YaleB and ORL databases. Following the comparison procedure, it is obvious that our methodology
exceeds the methods indicated above in terms of recognition rate. Thus, we may assume that our KMCNN has
the potential to be very effective in a wide range of computer vision applications.
4.3.2. Face recognition with noise
The second experiment was performed on the same databases, but instead of using original images,
we compared our model KMCNN with the proposed CNN architecture using noisy images. Each column of
Tables 6 to 11 illustrates the precision of the suggested model in terms of accuracy employing various salt and
pepper and speckle noise degradations, starting from 1% to 5%. Each row represents results obtained from
10. Int J Elec & Comp Eng ISSN: 2088-8708
Robust face recognition using convolutional neural networks combined … (Yassir El Madmoune)
4061
each order starting from lower orders to bigger ones, except the last row that shows the accuracy of the proposed
CNN architecture without using Krawtchouk moments.
According to the results shown in Tables 6, 8, and 10, the KMCNN obtained good classification rates
for various degradations of salt and pepper noise, beginning at order 40 when evaluated in YaleB, particularly
when the accuracy was 88.93% even with 5% of noise. We also remark that results of the KMCNN are more
accurate than those of CNN only, even if samples are under 5% of the same noise using ORL database. The
high robustness of the KMCNN can be noticed when we compare it with CNN taking (as input) noisy images
from LFW database; it achieves an accuracy between 71.58% and 73.97% compared with CNN that not even
surpassing 68% in terms of accuracy.
Taking into account the speckle noise classification rate values shown in Tables 7, 9, and 11, it can
be clearly observed that the KMCNN provides the highest classification rates with YaleB database, particularly
when the accuracy was 90.82% even with a variance of σ=0.4, and it performs better than CNN. By using noisy
images from ORL database the KMCNN surpasses 96%, while CNN only did not even surpass 93%. Alternatively,
we notice also that the KMCNN gives interesting accuracies in very low orders using LFW database.
Considering the results depicted in Tables 6 to 11, it is evident that the accuracy values increase with
the order of the moments up to an optimal order; after that it starts to decrease, but what is important is that the
best results are obtained in lower orders and they are better than the results obtained by CNN. From this fact,
we may deduce that the KMCNN is highly noise-tolerant, which is necessary for face recognition in noisy
environments. Hence the KMCNN confirms the fact presented in [42], indicating that face recognition using
CNN is intolerant to noisy conditions.
Table 6. Classification accuracy using Krawtchouk moments and YaleB database in noise-free and salt and
pepper noisy environment
Krawtchouk
moments + CNN
Free Noise
Salt and Pepper noise
1 % 2 % 3 % 4 % 5 %
Order Accuracy Accuracy Accuracy Accuracy Accuracy Accuracy
10 28.74 25.91 25.91 27.39 24.83 25.10
20 66.93 65.45 63.02 64.23 60.59 61.26
40 89.47 77.59 83.13 85.56 82.45 84.07
60 90.41 88.66 87.85 87.98 87.58 86.63
80 90.55 89.87 90.41 89.60 89.33 88.93
100 90.95 88.93 89.33 89.47 88.79 87.98
120 91.22 89.87 87.71 87.71 86.90 86.36
140 90.95 89.06 87.17 87.04 86.36 84.48
160 91.76 87.71 85.96 85.56 85.15 83.80
168 92.03 87.98 84.88 84.34 84.21 81.91
CNN 94.90 81.64 80.56 81.78 80.16 80.16
Table 7. Classification accuracy using Krawtchouk moments and YaleB database in noise-free and speckle
noisy environment
Krawtchouk moments +CNN Free Noise
Speckle noise
1 % 2 % 3 % 4 % 5 %
Order Accuracy Accuracy Accuracy Accuracy Accuracy Accuracy
10 28.74 28.60 28.07 29.28 27.93 29.28
20 66.93 68.28 66.35 65.72 65.58 67.47
40 89.47 89.74 89.20 88.93 88.66 86.90
60 90.41 68.63 90.01 86.90 88.79 89.47
80 90.55 91.09 91.63 89.47 86.77 89.60
100 90.95 91.22 90.01 88.52 90.28 89.33
120 91.22 91.63 90.41 81.51 90.82 90.28
140 90.95 91.09 89.87 90.41 87.31 90.55
160 91.76 90.68 90.41 89.74 90.55 90.41
168 92.03 90.55 89.87 90.41 88.79 90.14
CNN 94.90 93.65 87.58 90.41 82.32 91.22
In the last experiment, we compared our KMCNN model with other models based on CNN combined
with 2D orthogonal moments like Tchebichef moments [60], Hahn moments [61], Racah moments [62] and
Krawtchouk moments combined with pre-trained VGG16 model [59] using the same noisy conditions
presented in the previous experiment. The accuracy results of the noisy images from YaleB, ORL and LFW
databases for the KMCNN and prementioned models are respectively shown in Figures 10 to 15, a descriptive
legend is given in Figure 16.
12. Int J Elec & Comp Eng ISSN: 2088-8708
Robust face recognition using convolutional neural networks combined … (Yassir El Madmoune)
4063
Examining the given results in the aforementioned figures, the proposed KMCNN achieved the
greatest recognition performance for the four classifiers on the three datasets. In fact, the depicted graphs all
demonstrate the same general trend, where the recognition rate values increase by increasing the order of the
noisy image moments up to an optimal order, then start to decrease. The obtained results indicate that the
KMCNN offers a better strategy to handle noise compared to the combination of CNN with other 2D discrete
orthogonal moments. Perhaps this is due to our suggested KMCNN is able to accurately reflect global features
by employing discrete orthogonal polynomials with a near-zero redundancy measure in a feature set, as well
as their robustness against the effects of noise.
Comparing the results with an architecture that use Krawtchouk moments with VGG16 [59] as pre-
trained convolutional neural networks, the KMCNN gives interesting accuracies. This is probably due to the
flexibly of the proposed CNN to take different dimension as input layer, however using pre-trained CNNs like
VGG16 requires a fixed input shape which lead to the necessity of resizing the image moment and transform
it to RGB format. As a result, the capacity of our architecture to represent appropriate features for face
recognition was proved. Finally, based on the results depicted in Figures 10 to 15, the proposed KMCNN has
reached very satisfactory recognition accuracies, even in a noisy environment, also, it might have a great utility
in real-world applications against this type of noise.
Figure 10. Classification accuracy for different orders using 2D discrete orthogonal moments moments+CNN
and Krawtchouk+VGG16 in noisy conditions with salt and pepper and YaleB database
Figure 11. Classification accuracy for different orders using 2D discrete orthogonal moments+CNN and
Krawtchouk moments+VGG16 in noisy conditions with speckle and YaleB database
Figure 12. Classification accuracy for different orders using 2D discrete orthogonal moments+CNN and
Krawtchouk moments+VGG16 in noisy conditions with salt and pepper and ORL database
13. ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 13, No. 4, August 2023: 4052-4067
4064
Figure 13. Classification accuracy for different orders using 2D discrete orthogonal moments+CNN and
Krawtchouk moments+VGG16 in noisy conditions with speckle and ORL database
Figure 14. Classification accuracy for different using 2D discrete orthogonal moments +CNN and
Krawtchouk moments+VGG16 in noisy conditions with salt and pepper and LFW database
Figure 15. Classification accuracy for different orders using 2D discrete orthogonal moments+CNN and
Krawtchouk moments+VGG16 in noisy conditions with speckle and LFW database
Figure 16. A clear legend for Figures 10 to15 presented above
5. CONCLUSION
In this paper, we have suggested a novel face recognition approach that can tolerate deformations
produced by two forms of noise: salt and pepper and speckle. The suggested model is founded on the
combination of features extracted by the calculation of Krawtchouk moments and convolutional neural
networks. Applying Krawtchouk moments on images produced various feature vectors that were then fed into
CNN's input layer. The proposed model performed well on small-sized face images (70×80) from the ORL
database, large-sized face images (168×192) from the YaleB database, and images (240×240) from the LFW
database. The experimental results demonstrated that the suggested model enhanced the accuracy of face
recognition with noisy images and surpassed CNN alone and when we combined it with 2D discrete moments
14. Int J Elec & Comp Eng ISSN: 2088-8708
Robust face recognition using convolutional neural networks combined … (Yassir El Madmoune)
4065
like Tchebichef Hahn and Racah significantly. For future works, we plan to further examinate the robustness
of the proposed model using different types of noise. We also plan to extend our model to improve the accuracy
of 3D noisy face images.
ACKNOWLEDGEMENTS
The authors thankfully acknowledge the Laboratory of Intelligent Systems and Applications (LSIA)
for his support to achieve this work.
REFERENCES
[1] A. K. Jain, A. A. Ross, and K. Nandakumar, “Face recognition,” in Introduction to Biometrics, Boston, MA: Springer US, 2011,
pp. 97–139.
[2] A. Ono, “Face recognition with Zernike moments,” Systems and Computers in Japan, vol. 34, no. 10, pp. 26–35, Sep. 2003, doi:
10.1002/scj.10414.
[3] Y. V. Lata et al., “Facial recognition using eigenfaces by PCA,” International Journal of Recent Trends in Engineering, vol. 1,
no. 1, pp. 587–590, 2009.
[4] P. B. Balla and K. T. Jadhao, “IoT based facial recognition security system,” in 2018 International Conference on Smart City and
Emerging Technology (ICSCET), Jan. 2018, pp. 1–4, doi: 10.1109/ICSCET.2018.8537344.
[5] L. M. Mayron, “Biometric authentication on mobile devices,” IEEE Security & Privacy, vol. 13, no. 3, pp. 70–73, May 2015, doi:
10.1109/MSP.2015.67.
[6] M. Owayjan, A. Dergham, G. Haber, N. Fakih, A. Hamoush, and E. Abdo, “Face recognition security system,” in New Trends in
Networking, Computing, E-learning, Systems Sciences, and Engineering, 2015, pp. 343–348.
[7] D.-L. Wu, W. W. Y. Ng, P. P. K. Chan, H.-L. Ding, B.-Z. Jing, and D. S. Yeung, “Access control by RFID and face recognition
based on neural network,” in 2010 International Conference on Machine Learning and Cybernetics, Jul. 2010, pp. 675–680, doi:
10.1109/ICMLC.2010.5580558.
[8] J. R. Barr, K. W. Bowyer, P. J. Flynn, and S. Biswas, “Face recognition from video: a review,” International Journal of Pattern
Recognition and Artificial Intelligence, vol. 26, no. 05, Aug. 2012, doi: 10.1142/S0218001412660024.
[9] M. C. de Pinho, N. M. Ribeiro, and F. R. Gouveia, “Automatic detection of human faces in uncontrolled environments: Identification
of direction and movement,” in 6th Iberian Conference on Information Systems and Technologies (CISTI 2011), 2011, pp. 1–7.
[10] G. B. Huang, M. Ramesh, T. Berg, and E. Learned-Miller, “Labeled faces in the wild: A database for studying face recognition in
unconstrained environments,” in Workshop on faces in’Real-Life’Images: detection, alignment, and recognition, 2008, pp. 1–11.
[11] M. Turk and A. Pentland, “Eigenfaces for recognition,” Journal of Cognitive Neuroscience, vol. 3, no. 1, pp. 71–86, Jan. 1991, doi:
10.1162/jocn.1991.3.1.71.
[12] H. J. Seo and P. Milanfar, “Face verification using the LARK representation,” IEEE Transactions on Information Forensics and
Security, vol. 6, no. 4, pp. 1275–1286, Dec. 2011, doi: 10.1109/TIFS.2011.2159205.
[13] K. Simonyan, O. Parkhi, A. Vedaldi, and A. Zisserman, “Fisher vector faces in the wild,” in Proceedings of the British Machine
Vision Conference 2013, 2013, pp. 8.1--8.11, doi: 10.5244/C.27.8.
[14] B. Li and K.-K. Ma, “Fisherface vs. Eigenface in the dual-tree complex wavelet domain,” in 2009 Fifth International Conference
on Intelligent Information Hiding and Multimedia Signal Processing, Sep. 2009, pp. 30–33, doi: 10.1109/IIH-MSP.2009.322.
[15] M. Annalakshmi, S. M. M. Roomi, and A. S. Naveedh, “A hybrid technique for gender classification with SLBP and HOG features,”
Cluster Computing, vol. 22, no. S1, pp. 11–20, Jan. 2019, doi: 10.1007/s10586-017-1585-x.
[16] Z.-H. Huang, W.-J. Li, J. Shang, J. Wang, and T. Zhang, “Non-uniform patch based face recognition via 2D-DWT,” Image and
Vision Computing, vol. 37, pp. 12–19, May 2015, doi: 10.1016/j.imavis.2014.12.005.
[17] Z. Sufyanu, F. S. Mohamad, A. A. Yusuf, and M. B. Mamat, “Enhanced face recognition using discrete cosine transform,”
Engineering Letters, vol. 24, no. 1, pp. 52–61, 2016.
[18] J. H. Shah, M. Sharif, M. Raza, and A. Azeem, “A survey: Linear and nonlinear PCA based face recognition techniques,”
International Arab Journal of Information Technology, vol. 10, no. 6, pp. 536–545, 2013.
[19] S. R. Arashloo and J. Kittler, “Class-specific kernel fusion of multiple descriptors for face verification using multiscale binarised
statistical image features,” IEEE Transactions on Information Forensics and Security, vol. 9, no. 12, pp. 2100–2109, Dec. 2014,
doi: 10.1109/TIFS.2014.2359587.
[20] A. Vinay, V. S. Shekhar, K. N. B. Murthy, and S. Natarajan, “Performance study of LDA and KFA for Gabor based face recognition
system,” Procedia Computer Science, vol. 57, pp. 960–969, 2015, doi: 10.1016/j.procs.2015.07.493.
[21] S. Lawrence, C. L. Giles, Ah Chung Tsoi, and A. D. Back, “Face recognition: a convolutional neural-network approach,” IEEE
Transactions on Neural Networks, vol. 8, no. 1, pp. 98–113, Jan. 1997, doi: 10.1109/72.554195.
[22] P. Khoi, L. Huu, and V. Hoai, “Face retrieval based on local binary pattern and its variants: A comprehensive study,” International
Journal of Advanced Computer Science and Applications, vol. 7, no. 6, pp. 249–258, 2016, doi: 10.14569/IJACSA.2016.070632.
[23] M. Karaaba, O. Surinta, L. Schomaker, and M. A. Wiering, “Robust face recognition by computing distances from multiple
histograms of oriented gradients,” in 2015 IEEE Symposium Series on Computational Intelligence, Dec. 2015, pp. 203–209, doi:
10.1109/SSCI.2015.39.
[24] C. S. Weaver and J. W. Goodman, “A technique for optically convolving two functions,” Applied Optics, vol. 5, no. 7,
pp. 1248–1249, Jul. 1966, doi: 10.1364/AO.5.001248.
[25] A. V Lugt, “Signal detection by complex spatial filtering,” IEEE Transactions on Information Theory, vol. 10, no. 2, pp. 139–145,
Apr. 1964, doi: 10.1109/TIT.1964.1053650.
[26] J. S. Rani, D. Devaraj, and R. Sukanesh, “A novel feature extraction technique for face recognition,” in International Conference
on Computational Intelligence and Multimedia Applications (ICCIMA 2007), 2007, pp. 428–435, doi: 10.1109/ICCIMA.2007.141.
[27] L. Lenc and P. Král, “Automatic face recognition system based on the SIFT features,” Computers & Electrical Engineering, vol.
46, pp. 256–272, Aug. 2015, doi: 10.1016/j.compeleceng.2015.01.014.
[28] G. Du, F. Su, and A. Cai, “Face recognition using SURF features,” in MIPPR 2009: Pattern Recognition and Computer Vision,
Oct. 2009, pp. 593–599, doi: 10.1117/12.832636.
[29] M. Calonder, V. Lepetit, M. Ozuysal, T. Trzcinski, C. Strecha, and P. Fua, “BRIEF: Computing a local binary descriptor very fast,”
IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 34, no. 7, pp. 1281–1298, Jul. 2012, doi:
10.1109/TPAMI.2011.222.
15. ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 13, No. 4, August 2023: 4052-4067
4066
[30] A. Alahi, R. Ortiz, and P. Vandergheynst, “FREAK: Fast retina keypoint,” in 2012 IEEE Conference on Computer Vision and
Pattern Recognition, Jun. 2012, pp. 510–517, doi: 10.1109/CVPR.2012.6247715.
[31] A. A. Fathima, S. Ajitha, V. Vaidehi, M. Hemalatha, R. Karthigaiveni, and R. Kumar, “Hybrid approach for face recognition
combining Gabor Wavelet and Linear Discriminant Analysis,” in 2015 IEEE International Conference on Computer Graphics,
Vision and Information Security (CGVIS), Nov. 2015, pp. 220–225, doi: 10.1109/CGVIS.2015.7449925.
[32] Y. Yan, H. Wang, and D. Suter, “Multi-subregion based correlation filter bank for robust face recognition,” Pattern Recognition,
vol. 47, no. 11, pp. 3487–3501, Nov. 2014, doi: 10.1016/j.patcog.2014.05.004.
[33] C. Ding and D. Tao, “Robust face recognition via multimodal deep face representation,” IEEE Transactions on Multimedia, vol.
17, no. 11, pp. 2049–2058, Nov. 2015, doi: 10.1109/TMM.2015.2477042.
[34] F. Juefei-Xu, K. Luu, and M. Savvides, “Spartans: Single-sample periocular-based alignment-robust recognition technique applied
to non-frontal scenarios,” IEEE Transactions on Image Processing, vol. 24, no. 12, pp. 4780–4795, Dec. 2015, doi:
10.1109/TIP.2015.2468173.
[35] Y. Ding, Y. Cheng, X. Cheng, B. Li, X. You, and X. Yuan, “Noise-resistant network: a deep-learning method for face recognition
under noise,” EURASIP Journal on Image and Video Processing, vol. 2017, no. 1, Dec. 2017, doi: 10.1186/s13640-017-0188-z.
[36] X. Meng, Y. Yan, S. Chen, and H. Wang, “A cascaded noise-robust deep CNN for face recognition,” in 2019 IEEE International
Conference on Image Processing (ICIP), Sep. 2019, pp. 3487–3491, doi: 10.1109/ICIP.2019.8803443.
[37] X. Wu, R. He, Z. Sun, and T. Tan, “A light CNN for deep face representation with noisy labels,” IEEE Transactions on Information
Forensics and Security, vol. 13, no. 11, pp. 2884–2896, Nov. 2018, doi: 10.1109/TIFS.2018.2833032.
[38] Z. Ma, Y. Ding, B. Li, and X. Yuan, “Deep CNNs with robust LBP guiding pooling for face recognition,” Sensors, vol. 18, no. 11,
Nov. 2018, doi: 10.3390/s18113876.
[39] T. S. Arulananth, B. Manjula, and M. Baskar, “Human position tracking and detection using geometric active contours,” in 2020
Second International Conference on Inventive Research in Computing Applications (ICIRCA), Jul. 2020, pp. 509–512, doi:
10.1109/ICIRCA48905.2020.9182825.
[40] M. Jahangir Alam, T. Chowdhury, and M. Shahzahan Ali, “A smart login system using face detection and recognition by ORB
algorithm,” Indonesian Journal of Electrical Engineering and Computer Science (IJEECS), vol. 20, no. 2, pp. 1078–1087, Nov.
2020, doi: 10.11591/ijeecs.v20.i2.pp1078-1087.
[41] D. Rolnick, A. Veit, S. Belongie, and N. Shavit, “Deep learning is robust to massive label noise,” Prepr. arXiv.1705.10694, 2017.
[42] F. Wang et al., “The devil of face recognition is in the noise,” in Computer Vision – ECCV 2018, 2018, pp. 780–795.
[43] Pew-Thian Yap, R. Paramesran, and Seng-Huat Ong, “Image analysis by Krawtchouk moments,” IEEE Transactions on Image
Processing, vol. 12, no. 11, pp. 1367–1377, Nov. 2003, doi: 10.1109/TIP.2003.818019.
[44] K. D. Apostolidis and G. A. Papakostas, “Digital watermarking as an adversarial attack on medical image analysis with deep
learning,” Journal of Imaging, vol. 8, no. 6, May 2022, doi: 10.3390/jimaging8060155.
[45] H. Amakdouf, A. Zouhri, M. EL Mallahi, and H. Qjidaa, “Color image analysis of quaternion discrete radial Krawtchouk moments,”
Multimedia Tools and Applications, vol. 79, no. 35–36, pp. 26571–26586, Sep. 2020, doi: 10.1007/s11042-020-09120-0.
[46] G. Hassan, K. M. Hosny, R. M. Farouk, and A. M. Alzohairy, “New set of invariant quaternion Krawtchouk moments for color
image representation and recognition,” International Journal of Image and Graphics, vol. 22, no. 04, Jul. 2022, doi:
10.1142/S0219467822500371.
[47] S. M. M. Rahman, T. Howlader, and D. Hatzinakos, “On the selection of 2D Krawtchouk moments for face recognition,” Pattern
Recognition, vol. 54, pp. 83–93, Jun. 2016, doi: 10.1016/j.patcog.2016.01.003.
[48] M. Krawtchouk, “On interpolation by means of orthogonal polynomials,” Memoirs Agricultural Inst. Kyiv, vol. 4, pp. 21–28, 1929.
[49] P. T. Yap, P. Raveendran, and S. H. Ong, “Krawtchouk moments as a new set of discrete orthogonal moments for image
reconstruction,” in Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN’02 (Cat. No.02CH37290),
2002, pp. 908–912, doi: 10.1109/IJCNN.2002.1005595.
[50] R. Mukundan, P. Raveendran, and W. A. Jassim, “New orthogonal polynomials for speech signal and image processing,” IET Signal
Processing, vol. 6, no. 8, pp. 713–723, Oct. 2012, doi: 10.1049/iet-spr.2011.0004.
[51] K.-H. Thung, R. Paramesran, and C.-L. Lim, “Content-based image quality metric using similarity measure of moment vectors,”
Pattern Recognition, vol. 45, no. 6, pp. 2193–2204, Jun. 2012, doi: 10.1016/j.patcog.2011.12.001.
[52] S. H. Abdulhussain, A. R. Ramli, S. A. R. Al-Haddad, B. M. Mahmmod, and W.A. Jassim, “On computational aspects of Tchebichef
polynomials for higher polynomial order,” IEEE Access, vol. 5, pp. 2470–2478, 2017, doi: 10.1109/ACCESS.2017.2669218.
[53] K.-C. Lee, J. Ho, and D. J. Kriegman, “Acquiring linear subspaces for face recognition under variable lighting,” IEEE Transactions
on Pattern Analysis and Machine Intelligence, vol. 27, no. 5, pp. 684–698, May 2005, doi: 10.1109/TPAMI.2005.92.
[54] V. Nair and G. E. Hinton, “Rectified linear units improve restricted Boltzmann machines,” in ICML’10: Proceedings of the 27th
International Conference on International Conference on Machine Learning, 2010, pp. 807–814.
[55] B. Xu, N. Wang, T. Chen, and M. Li, “Empirical evaluation of rectified activations in convolutional network,” Prepr.
arXiv.1505.00853, 2015.
[56] X. Glorot and Y. Bengio, “Understanding the difficulty of training deep feedforward neural networks,” in Proceedings of the
thirteenth international conference on artificial intelligence and statistics, 2010, pp. 249–256.
[57] S. Ioffe and C. Szegedy, “Batch normalization: accelerating deep network training by reducing internal covariate shift,” in 32nd
International Conference on Machine Learning, ICML 2015, 2015, vol. 1, pp. 448–456.
[58] “The database of faces,” AT&T Laboratories Cambridge. 2001, Accessed: Mar. 10, 2022. [Online]. Available: https://cam-
orl.co.uk/facedatabase.html.
[59] K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” arXiv preprint
arXiv:1409.1556, Sep. 2014.
[60] R. Mukundan, S. H. Ong, and P. A. Lee, “Image analysis by tchebichef moments,” IEEE Transactions on Image Processing,
vol. 10, no. 9, pp. 1357–1364, 2001, doi: 10.1109/83.941859.
[61] P.-T. Yap, R. Paramesran, and S.-H. Ong, “Image analysis using Hahn moments,” IEEE Transactions on Pattern Analysis and
Machine Intelligence, vol. 29, no. 11, pp. 2057–2062, Nov. 2007, doi: 10.1109/TPAMI.2007.70709.
[62] H. Zhu, H. Shu, J. Liang, L. Luo, and J.-L. Coatrieux, “Image analysis by discrete orthogonal Racah moments,” Signal Processing,
vol. 87, no. 4, pp. 687–708, Apr. 2007, doi: 10.1016/j.sigpro.2006.07.007.
[63] Y. Jiang, Y. Wu, W. Li, L. Wang, and Q. Liao, “Log-domain polynomial filters for illumination-robust face recognition,” in 2014
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), May 2014, pp. 504–508, doi:
10.1109/ICASSP.2014.6853647.
[64] Ngoc-Son Vu and A. Caplier, “Enhanced patterns of oriented edge magnitudes for face recognition and image matching,” IEEE
Transactions on Image Processing, vol. 21, no. 3, pp. 1352–1365, Mar. 2012, doi: 10.1109/TIP.2011.2166974.
16. Int J Elec & Comp Eng ISSN: 2088-8708
Robust face recognition using convolutional neural networks combined … (Yassir El Madmoune)
4067
[65] M. M. Abdelwahab, S. A. Aly, and I. Yousry, “Efficient web-based facial recognition system employing 2DHOG,” Prepr.
arXiv.1202.2449, 2012.
[66] Y. Wang, Z. Xu, W. Li, and Q. Liao, “Illumination-robust face recognition with block-based local contrast patterns,” in 2017 IEEE
International Conference on Acoustics, Speech and Signal Processing (ICASSP), Mar. 2017, pp. 1418–1422, doi:
10.1109/ICASSP.2017.7952390.
[67] Y. Feng, X. An, and X. Liu, “The application of scale invariant feature transform fused with shape model in the human face
recognition,” in 2016 IEEE Advanced Information Management, Communicates, Electronic and Automation Control Conference
(IMCEC), Oct. 2016, pp. 1716–1720, doi: 10.1109/IMCEC.2016.7867511.
[68] Y. Gan, T. Yang, and C. He, “A deep graph embedding network model for face recognition,” in 2014 12th International Conference
on Signal Processing (ICSP), Oct. 2014, pp. 1268–1271, doi: 10.1109/ICOSP.2014.7015203.
[69] A. Vinay, D. Hebbar, V. S. Shekhar, K. N. B. Murthy, and S. Natarajan, “Two novel detector-descriptor based approaches for face
recognition using SIFT and SURF,” Procedia Computer Science, vol. 70, pp. 185–197, 2015, doi: 10.1016/j.procs.2015.10.070.
[70] T. M. Abhishree, J. Latha, K. Manikantan, and S. Ramachandran, “Face recognition using Gabor filter based feature extraction with
anisotropic diffusion as a pre-processing technique,” Procedia Computer Science, vol. 45, pp. 312–321, 2015, doi:
10.1016/j.procs.2015.03.149.
[71] K. Chaturvedi and D. K. Vishwakarma, “Face recognition in an unconstrained environment using ConvNet,” in Proceedings of the
2020 2nd International Conference on Big Data Engineering and Technology, 2020, pp. 67–71, doi: 10.1145/3378904.3378905.
BIOGRAPHIES OF AUTHORS
Yassir El Madmoune He received a M.S. degree in Intelligent Systems and
Networks, from the Faculty of Science and Technology, University of Sidi Mohammed Ben
Abdellah, Fez, Morocco in 2018. He is currently pursuing his Ph.D. degree in Computer Science
at the Faculty of Science and Technology of Fez. His research interests include pattern
recognition and computer vision. Email: yassir.elmadmoune@usmba.ac.ma.
Ilham El Ouariachi She received a M.S. degree in Intelligent Systems and
Networks, from the Faculty of Science and Technology, University of Sidi Mohammed Ben
Abdellah, Fez, Morocco in 2016. He is currently pursuing his Ph.D. degree in Computer Science
at the Faculty of Science and Technology of Fez. His research interests include pattern
recognition and computer vision. Email: ilham.elouariachi@usmba.ac.ma.
Khalid Zenkouar He received a Ph.D. degree in image analysis from Faculty of
Science, University Sidi Mohamed Ben Abdellah, Fez, Morocco in 2006. Now he is a professor
of the Department of computer engineering, Faculty of Science and Technology Fez Morocco.
He is a member in the LSIA Laboratory (Laboratory of Intelligent Systems and Application).
His current research interests include image analysis, machine intelligence and pattern
recognition. Email: khalid.zenkouar@usmba.ac.ma.
Azeddine Zahi Received his PhD degree in 1997 in Computer Sciences from
Mohammed V University in Rabat. He is a research professor at Sidi Mohamed Ben Abbdellah
University of Fez since 1995. His research interests include fuzzy data mining, heuristics and
automatic learning. Email: azeddine.zahi@usmba.ac.ma.